Datasets:

Hebrish
/

AI-Model-Sentences

id string	text string	duration_s float64	categories list	target_terms list	prompt_id string	prompt_source string	mode string	edited bool	device string	environment string	sample_rate int64	recorded_at string	microphone string	prompt_origin string	hebrew_pronunciation string	variant_groups list	variant_spelling string	speaker_id string	conditions list	imported_from string	original_text string	text_before_normalization string	text_he string	noise_level string	background_speech_languages list	notes string	model_hypothesis string	asr_model string	speaker string	speaker_gender string	noise_flags list	clipped bool	task string	spoken_text string	aside_text string	aside_position string	verbatim_text string	prompt_categories list	source_prompt_id string	prompt_kind string	prompt_note string	term_language string	paragraph_text string	contributor string	bit_depth int64	audio_format string	native_rate int64	capture_mode string	tts_eligible bool	tts_split string	transcript_status string	prompt_set string	topic string	style string	style_instruction string	written_form_uncertain bool	session_id string	room_tone_id string	peak_dbfs float64	floor_dbfs float64	speech_dbfs float64	snr_db float64	edge_head_dbfs float64	edge_tail_dbfs float64	browser_processing bool	room string	priority_word string	location string	location_name string	location_note string	latitude float64	longitude float64	location_accuracy_m float64	text_full string	fallback_ok bool	sentence_id string	pronunciation string	extra string
20261101T060615Z_gxa890	Switched the drafts over to GPT-6.1 Terra Pro this morning, just to compare.	6.037	[ "ai-models" ]	[ "GPT-6.1 Terra Pro" ]	ai-models_af58c5b5df	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-01T06:06:15.671Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Switched the drafts over to GPT-6.1 Terra Pro this morning, just to compare.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Switched the drafts over to GPT-6.1 Terra Pro this morning, just to compare.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-6.7	-53.6	-24.2	29.4	-37.1	-38.6	false	null	null	null	null	null	null	null	null	null	null	p_5d3a25a8e4	[]	{"pronunciation_variants": [], "ui_note": null}
20261101T061132Z_nr7qcx	Note to self: use GPT-6.1 Terra Pro for the client summaries, not the cheaper one.	6.738	[ "ai-models" ]	[ "GPT-6.1 Terra Pro" ]	ai-models_ac6735200c	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-01T06:11:32.457Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Note to self: use GPT-6.1 Terra Pro for the client summaries, not the cheaper one.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Note to self: use GPT-6.1 Terra Pro for the client summaries, not the cheaper one.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-5	-47.6	-17.7	29.8	-34.9	-38.6	false	null	null	null	null	null	null	null	null	null	null	p_f250ca8ce8	[]	{"pronunciation_variants": [], "ui_note": null}
20261101T061403Z_26mp14	Hey, did you try GPT-6.1 Terra yet? It handled my messy meeting notes better than I expected.	6.922	[ "ai-models" ]	[ "GPT-6.1 Terra" ]	ai-models_359dcad5d9	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-01T06:14:03.762Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Hey, did you try GPT-6.1 Terra yet? It handled my messy meeting notes better than I expected.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Hey, did you try GPT-6.1 Terra yet? It handled my messy meeting notes better than I expected.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-4.2	-51.2	-19.4	31.8	-39.7	-36.9	false	null	null	null	null	null	null	null	null	null	null	p_db0e38d60e	[]	{"pronunciation_variants": [], "ui_note": null}
20261101T061944Z_cx7y4g	Quick email to Dara: the summary came from GPT-6.1 Terra, so please check the figures before sending.	7.701	[ "ai-models" ]	[ "GPT-6.1 Terra" ]	ai-models_a73c0e08af	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-01T06:19:44.263Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Quick email to Dara: the summary came from GPT-6.1 Terra, so please check the figures before sending.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Quick email to Dara: the summary came from GPT-6.1 Terra, so please check the figures before sending.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-5	-48.2	-19	29.2	-36.8	-37.4	false	null	null	null	null	null	null	null	null	null	null	p_7afea092a6	[]	{"pronunciation_variants": [], "ui_note": null}
20261101T062451Z_1n9bpd	Switching the transcription notes over to GPT-6.1 Sol since it is quicker on short prompts.	6.93	[ "ai-models" ]	[ "GPT-6.1 Sol" ]	ai-models_0cbbc12741	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-01T06:24:51.165Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Switching the transcription notes over to GPT-6.1 Sol since it is quicker on short prompts.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Switching the transcription notes over to GPT-6.1 Sol since it is quicker on short prompts.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-4.5	-48	-20.1	27.9	-37.5	-34.8	false	null	null	null	null	null	null	null	null	null	null	p_b0c322c6df	[]	{"pronunciation_variants": [], "ui_note": null}
20261101T062609Z_i9cxuh	Reminder: the default is GPT-6.1 Sol, so change it if you need more depth.	6.223	[ "ai-models" ]	[ "GPT-6.1 Sol" ]	ai-models_c8192f78ae	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-01T06:26:09.642Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Reminder: the default is GPT-6.1 Sol, so change it if you need more depth.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Reminder: the default is GPT-6.1 Sol, so change it if you need more depth.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-2.8	-53.6	-18.4	35.2	-44.7	-33.8	false	null	null	null	null	null	null	null	null	null	null	p_96cb0d511a	[]	{"pronunciation_variants": [], "ui_note": null}
20261101T062920Z_r4a1dx	Ran the same brief through GPT-6 Luna Pro and the answer came back in a neat list.	6.479	[ "ai-models" ]	[ "GPT-6 Luna Pro" ]	ai-models_8ab02fdfc6	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-01T06:29:20.141Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Ran the same brief through GPT-6 Luna Pro and the answer came back in a neat list.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Ran the same brief through GPT-6 Luna Pro and the answer came back in a neat list.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-4.4	-52.3	-21.5	30.8	-40.8	-41.7	false	null	null	null	null	null	null	null	null	null	null	p_e47148f945	[]	{"pronunciation_variants": [], "ui_note": null}
20261101T063407Z_nt7xix	Would GPT-6 Luna Pro be fine for the weekly report, or should I stick with the bigger one?	5.989	[ "ai-models" ]	[ "GPT-6 Luna Pro" ]	ai-models_be61115d9c	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-01T06:34:07.843Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Would GPT-6 Luna Pro be fine for the weekly report, or should I stick with the bigger one?	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Would GPT-6 Luna Pro be fine for the weekly report, or should I stick with the bigger one?	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-10.2	-51.7	-21.3	30.4	-36.9	-40	false	null	null	null	null	null	null	null	null	null	null	p_70df976011	[]	{"pronunciation_variants": [], "ui_note": null}
20261101T063535Z_a7dqcd	Using GPT-6 Luna for the quick replies now, mainly because it keeps things short and clear.	6.33	[ "ai-models" ]	[ "GPT-6 Luna" ]	ai-models_5818806a70	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-01T06:35:35.858Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Using GPT-6 Luna for the quick replies now, mainly because it keeps things short and clear.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Using GPT-6 Luna for the quick replies now, mainly because it keeps things short and clear.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-4.8	-47.2	-22	25.2	-38.2	-35.6	false	null	null	null	null	null	null	null	null	null	null	p_bbda56328c	[]	{"pronunciation_variants": [], "ui_note": null}
20261101T063906Z_uumbeg	Told the team that GPT-6 Luna wrote the first draft, so they know to read it closely.	5.834	[ "ai-models" ]	[ "GPT-6 Luna" ]	ai-models_95987e61eb	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-01T06:39:06.212Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Told the team that GPT-6 Luna wrote the first draft, so they know to read it closely.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Told the team that GPT-6 Luna wrote the first draft, so they know to read it closely.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-6.8	-49.7	-20.1	29.6	-28.4	-24.8	false	null	null	null	null	null	null	null	null	null	null	p_7a32f0389b	[]	{"pronunciation_variants": [], "ui_note": null}
20261101T064748Z_lez4yl	Morning, GPT-6 Sol has been solid on my spelling checks so far.	5.045	[ "ai-models" ]	[ "GPT-6 Sol" ]	ai-models_770d3d15df	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-01T06:47:48.385Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Morning, GPT-6 Sol has been solid on my spelling checks so far.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Morning, GPT-6 Sol has been solid on my spelling checks so far.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-3.1	-48.5	-17.1	31.4	-38.2	-38.8	false	null	null	null	null	null	null	null	null	null	null	p_6d5aa07ce1	[]	{"pronunciation_variants": [], "ui_note": null}
20261101T070549Z_k9env7	Can we keep GPT-6 Sol for the proposal, or is the cost too high for a long document?	5.463	[ "ai-models" ]	[ "GPT-6 Sol" ]	ai-models_ee7e4303c1	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-01T07:05:49.319Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Can we keep GPT-6 Sol for the proposal, or is the cost too high for a long document?	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Can we keep GPT-6 Sol for the proposal, or is the cost too high for a long document?	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-3.9	-42.1	-18.7	23.3	-15.7	-32.3	false	null	null	null	null	null	null	null	null	null	null	p_31054b7475	[]	{"pronunciation_variants": [], "ui_note": null}
20261101T073415Z_dwzrgv	I set GPT-6 Astra as the default for the coding help, and it has behaved so far.	6.442	[ "ai-models" ]	[ "GPT-6 Astra" ]	ai-models_68250b9a3f	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-01T07:34:15.099Z	internal	dataset	null	[]	null	8992	[]	null	null	null	I set GPT-6 Astra as the default for the coding help, and it has behaved so far.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	I set GPT-6 Astra as the default for the coding help, and it has behaved so far.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-1.6	-27.3	-14.3	13	-13.6	-17.7	false	null	null	null	null	null	null	null	null	null	null	p_a3e1b80466	[]	{"pronunciation_variants": [], "ui_note": null}
20261101T074041Z_urssj7	Before the Thursday call, test GPT-6 Astra against the old prompts.	4.826	[ "ai-models" ]	[ "GPT-6 Astra" ]	ai-models_a6a0009d56	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-01T07:40:41.505Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Before the Thursday call, test GPT-6 Astra against the old prompts.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Before the Thursday call, test GPT-6 Astra against the old prompts.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-1.9	-38	-14.4	23.7	-24.8	-27.9	false	null	null	null	null	null	null	null	null	null	null	p_7ec38460bb	[]	{"pronunciation_variants": [], "ui_note": null}
20261101T075906Z_nvnecz	Can you switch the draft over to GPT-5.6 Terra and see if the tone improves?	5.679	[ "ai-models" ]	[ "GPT-5.6 Terra" ]	ai-models_3ba4ed3f64	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-01T07:59:06.018Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Can you switch the draft over to GPT-5.6 Terra and see if the tone improves?	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Can you switch the draft over to GPT-5.6 Terra and see if the tone improves?	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-1	-50.9	-10.3	40.5	-17.1	-56.5	false	null	null	null	null	null	null	null	null	null	null	p_2584103205	[]	{"pronunciation_variants": [], "ui_note": null}
20261101T081432Z_0qt0v7	Family chat, I used GPT-5.6 Terra for the shopping list and it was surprisingly sensible.	7.738	[ "ai-models" ]	[ "GPT-5.6 Terra" ]	ai-models_8a7b1dff4f	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T08:14:32.066Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Family chat, I used GPT-5.6 Terra for the shopping list and it was surprisingly sensible.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Family chat, I used GPT-5.6 Terra for the shopping list and it was surprisingly sensible.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-1	-30	-9.5	20.5	-15.6	-11.7	false	null	null	null	null	null	null	null	null	null	null	p_bab897e8ca	[]	{"pronunciation_variants": [], "ui_note": null, "activity": "walking"}
20261101T081512Z_6ab3zg	Note to self: try GPT-5.6 Sol on the longer prompts before Friday.	6.138	[ "ai-models" ]	[ "GPT-5.6 Sol" ]	ai-models_740bda835e	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T08:15:12.096Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Note to self: try GPT-5.6 Sol on the longer prompts before Friday.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Note to self: try GPT-5.6 Sol on the longer prompts before Friday.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-1	-69.3	-10.5	58.8	-56.9	-52.7	false	null	null	null	null	null	null	null	null	null	null	p_94633e6b4b	[]	{"pronunciation_variants": [], "ui_note": null, "activity": "walking"}
20261101T082313Z_9hqtf1	Is GPT-5.6 Luna fast enough for the live chat, or should we stick with the bigger one?	5.439	[ "ai-models" ]	[ "GPT-5.6 Luna" ]	ai-models_4356f7d8b6	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T08:23:13.366Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Is GPT-5.6 Luna fast enough for the live chat, or should we stick with the bigger one?	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Is GPT-5.6 Luna fast enough for the live chat, or should we stick with the bigger one?	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-6.2	-53.6	-20.9	32.7	-44.6	-45.3	false	null	null	null	null	null	null	null	null	null	null	p_346f8f3992	[]	{"pronunciation_variants": [], "ui_note": null, "activity": "walking"}
20261101T082427Z_eiah0c	I'd pick GPT-5.6 Luna for quick rewrites, though I haven't tested it much yet.	6.389	[ "ai-models" ]	[ "GPT-5.6 Luna" ]	ai-models_6266012b8b	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T08:24:27.851Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	I'd pick GPT-5.6 Luna for quick rewrites, though I haven't tested it much yet.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	I'd pick GPT-5.6 Luna for quick rewrites, though I haven't tested it much yet.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-6.5	-55.6	-22	33.6	-43.8	-43.1	false	null	null	null	null	null	null	null	null	null	null	p_38f7edfdc6	[]	{"pronunciation_variants": [], "ui_note": null, "activity": "walking"}
20261101T082503Z_wnw8nr	For the contract review I'd rather use GPT-5.5 Pro, even if it costs a bit more.	5.957	[ "ai-models" ]	[ "GPT-5.5 Pro" ]	ai-models_7aafeb70bd	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T08:25:03.875Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	For the contract review I'd rather use GPT-5.5 Pro, even if it costs a bit more.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	For the contract review I'd rather use GPT-5.5 Pro, even if it costs a bit more.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-6.3	-52.4	-20.4	32	-42.9	-42	false	null	null	null	null	null	null	null	null	null	null	p_a72d91155a	[]	{"pronunciation_variants": [], "ui_note": null, "activity": "walking"}
20261101T082523Z_z5koz8	Just a heads up, the notes from this morning came out of GPT-5.5 Pro, so double-check the dates.	6.693	[ "ai-models" ]	[ "GPT-5.5 Pro" ]	ai-models_d84a72de38	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T08:25:23.095Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Just a heads up, the notes from this morning came out of GPT-5.5 Pro, so double-check the dates.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Just a heads up, the notes from this morning came out of GPT-5.5 Pro, so double-check the dates.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-7.8	-51.3	-20.9	30.4	-43.2	-42.6	false	null	null	null	null	null	null	null	null	null	null	p_c97ec0b74a	[]	{"pronunciation_variants": [], "ui_note": null, "activity": "walking"}
20261101T082535Z_rqex1p	Switching the classroom bot to GPT-5.4 Mini should keep things quick and cheap, I reckon.	5.637	[ "ai-models" ]	[ "GPT-5.4 Mini" ]	ai-models_a24d9d1024	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T08:25:35.400Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Switching the classroom bot to GPT-5.4 Mini should keep things quick and cheap, I reckon.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Switching the classroom bot to GPT-5.4 Mini should keep things quick and cheap, I reckon.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-6.8	-52.7	-21.5	31.2	-41.5	-44.9	false	null	null	null	null	null	null	null	null	null	null	p_126d7811ac	[]	{"pronunciation_variants": [], "ui_note": null, "activity": "walking"}
20261101T083108Z_x6pk1f	Sorry for the typos, that reply was from GPT-5.4 Mini on my phone.	5.397	[ "ai-models" ]	[ "GPT-5.4 Mini" ]	ai-models_921b1a53e3	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T08:31:08.198Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Sorry for the typos, that reply was from GPT-5.4 Mini on my phone.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Sorry for the typos, that reply was from GPT-5.4 Mini on my phone.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261031T163313Z_jwfjq1	null	-4.3	-53.1	-19.9	33.1	-42.9	-43	false	null	null	null	null	null	null	null	null	null	null	p_6e1f879ec5	[]	{"pronunciation_variants": [], "ui_note": null, "activity": "walking"}
20261101T083726Z_cu6jmf	We could hand the tagging job to GPT-5.4 Nano and see whether it copes with the volume.	5.927	[ "ai-models" ]	[ "GPT-5.4 Nano" ]	ai-models_a1dfb40aca	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T08:37:26.483Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	We could hand the tagging job to GPT-5.4 Nano and see whether it copes with the volume.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	We could hand the tagging job to GPT-5.4 Nano and see whether it copes with the volume.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261101T083553Z_x7j3ui	null	-1	-70.1	-10.6	59.5	-53.2	-52.4	false	null	null	null	null	null	null	null	null	null	null	p_926c579918	[]	{"pronunciation_variants": [], "ui_note": null, "activity": "walking"}
20261101T083949Z_74yz2k	Quick question: did anyone already test GPT-5.4 Nano on the short product descriptions?	5.805	[ "ai-models" ]	[ "GPT-5.4 Nano" ]	ai-models_d4067e24c5	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T08:39:49.145Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Quick question: did anyone already test GPT-5.4 Nano on the short product descriptions?	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Quick question: did anyone already test GPT-5.4 Nano on the short product descriptions?	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261101T083553Z_x7j3ui	null	-1	-69.4	-10.8	58.6	-57.4	-56.9	false	null	null	null	null	null	null	null	null	null	null	p_186042957b	[]	{"pronunciation_variants": [], "ui_note": null, "activity": "walking"}
20261101T084054Z_te7f6m	I kept the original wording and only asked GPT-5.4 Pro to tidy the grammar.	5.77	[ "ai-models" ]	[ "GPT-5.4 Pro" ]	ai-models_3dccf8384b	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T08:40:54.138Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	I kept the original wording and only asked GPT-5.4 Pro to tidy the grammar.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	I kept the original wording and only asked GPT-5.4 Pro to tidy the grammar.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261101T083553Z_x7j3ui	null	-1	-69.5	-11	58.5	-57.2	-42.7	false	null	null	null	null	null	null	null	null	null	null	p_17a11b6a04	[]	{"pronunciation_variants": [], "ui_note": null, "activity": "walking"}
20261101T084110Z_v2ug5d	Family group, the birthday toast was drafted with GPT-5.4 Pro, so blame it if it's too formal.	7.935	[ "ai-models" ]	[ "GPT-5.4 Pro" ]	ai-models_e906fff794	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T08:41:10.220Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Family group, the birthday toast was drafted with GPT-5.4 Pro, so blame it if it's too formal.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Family group, the birthday toast was drafted with GPT-5.4 Pro, so blame it if it's too formal.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261101T083553Z_x7j3ui	null	-1	-69	-10.5	58.5	-45.6	-54.2	false	null	null	null	null	null	null	null	null	null	null	p_9ebb63abf5	[]	{"pronunciation_variants": [], "ui_note": null, "activity": "walking"}
20261101T085753Z_aawk5z	Open the repo, select GPT-5.2-Codex in the settings, and rerun the failing tests.	6.303	[ "ai-models" ]	[ "GPT-5.2-Codex" ]	ai-models_eb0a204c18	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T08:57:53.683Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Open the repo, select GPT-5.2-Codex in the settings, and rerun the failing tests.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Open the repo, select GPT-5.2-Codex in the settings, and rerun the failing tests.	null	null	null	null	null	coding model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261101T085338Z_todja1	null	-1	-69.7	-10.2	59.4	-57.8	-55.3	false	null	null	null	null	null	null	null	null	null	null	p_a505c1f041	[]	{"mic_link": null, "bt_codec": null, "activity": null, "pronunciation_variants": [], "ui_note": null}
20261101T090058Z_nub072	The refactor from yesterday came out of GPT-5.2-Codex, so read through the loops carefully.	6.357	[ "ai-models" ]	[ "GPT-5.2-Codex" ]	ai-models_617db7912c	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T09:00:58.633Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	The refactor from yesterday came out of GPT-5.2-Codex, so read through the loops carefully.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	The refactor from yesterday came out of GPT-5.2-Codex, so read through the loops carefully.	null	null	null	null	null	coding model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261101T085338Z_todja1	null	-1	-69.6	-9.7	60	-58.3	-55.5	false	null	null	null	null	null	null	null	null	null	null	p_71d762a7b7	[]	{"mic_link": null, "bt_codec": null, "activity": null, "pronunciation_variants": [], "ui_note": null}
20261101T091354Z_s3gldv	Planning the week, I think I'll try MiMo-V2.5 for the long document summaries before deciding.	5.994	[ "ai-models" ]	[ "MiMo-V2.5" ]	ai-models_1a13329818	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T09:13:54.854Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Planning the week, I think I'll try MiMo-V2.5 for the long document summaries before deciding.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Planning the week, I think I'll try MiMo-V2.5 for the long document summaries before deciding.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261101T091209Z_w7iep5	null	-1	-69.1	-10.5	58.6	-57.8	-52.9	false	null	null	null	null	null	null	null	null	null	null	p_ce4b8a6fda	[]	{"mic_link": null, "bt_codec": null, "activity": null, "pronunciation_variants": [], "ui_note": null}
20261101T091728Z_o7airk	The draft from MiMo-V2.5 reads well, so I'll forward it to the client this afternoon.	5.717	[ "ai-models" ]	[ "MiMo-V2.5" ]	ai-models_e195411872	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T09:17:28.490Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	The draft from MiMo-V2.5 reads well, so I'll forward it to the client this afternoon.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	The draft from MiMo-V2.5 reads well, so I'll forward it to the client this afternoon.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261101T091209Z_w7iep5	null	-1	-69.6	-9.7	59.9	-56.5	-54.6	false	null	null	null	null	null	null	null	null	null	null	p_dfce190177	[]	{"mic_link": null, "bt_codec": null, "activity": null, "pronunciation_variants": [], "ui_note": null}
20261101T092355Z_kn5rah	Quick one for you, I switched the support bot over to Nova 2 Lite because it's quicker.	5.509	[ "ai-models" ]	[ "Nova 2 Lite" ]	ai-models_f084d3b3d7	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T09:23:55.063Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Quick one for you, I switched the support bot over to Nova 2 Lite because it's quicker.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Quick one for you, I switched the support bot over to Nova 2 Lite because it's quicker.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261101T092057Z_cc5t9u	null	-1	-70.1	-10	60.1	-58.3	-57.8	false	null	null	null	null	null	null	null	null	null	null	p_d49ce39a82	[]	{"mic_link": null, "bt_codec": null, "activity": null, "pronunciation_variants": [], "ui_note": null}
20261101T092531Z_x7z8ak	Honestly, Nova 2 Lite is fine for tagging meeting notes, so let's stop overpaying.	5.653	[ "ai-models" ]	[ "Nova 2 Lite" ]	ai-models_d0c298b9d2	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T09:25:31.668Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Honestly, Nova 2 Lite is fine for tagging meeting notes, so let's stop overpaying.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Honestly, Nova 2 Lite is fine for tagging meeting notes, so let's stop overpaying.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261101T092057Z_cc5t9u	null	-1	-70.1	-10.4	59.7	-58.3	-55.2	false	null	null	null	null	null	null	null	null	null	null	p_1acc1ea436	[]	{"mic_link": null, "bt_codec": null, "activity": null, "pronunciation_variants": [], "ui_note": null}
20261101T092900Z_1z3wsg	Could you check whether Nova Premier handled the spreadsheet formulas properly last night?	5.189	[ "ai-models" ]	[ "Nova Premier" ]	ai-models_82065db1a9	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T09:29:00.937Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Could you check whether Nova Premier handled the spreadsheet formulas properly last night?	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Could you check whether Nova Premier handled the spreadsheet formulas properly last night?	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261101T092752Z_dcimm5	null	-1	-69.7	-10.1	59.6	-57.6	-56.3	false	null	null	null	null	null	null	null	null	null	null	p_fb9f53e656	[]	{"mic_link": null, "bt_codec": null, "activity": "seated", "pronunciation_variants": [], "ui_note": null}
20261101T094831Z_xhbphu	I've been comparing Nova Premier with my usual model all morning and the answers differ a lot.	6.357	[ "ai-models" ]	[ "Nova Premier" ]	ai-models_6b0436e1c6	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T09:48:31.459Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	I've been comparing Nova Premier with my usual model all morning and the answers differ a lot.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	I've been comparing Nova Premier with my usual model all morning and the answers differ a lot.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261101T093718Z_mw6z6o	null	-1	-71.3	-10	61.3	-58.3	-58.7	false	null	null	null	null	null	null	null	null	null	null	p_9928bf956f	[]	{"mic_link": null, "bt_codec": null, "activity": "seated", "pronunciation_variants": [], "ui_note": null}
20261101T095419Z_0ha5j0	LFM2.5 is small enough to run on my laptop, handy when the wifi drops.	5.509	[ "ai-models" ]	[ "LFM2.5" ]	ai-models_a448eb8593	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T09:54:19.718Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	LFM2.5 is small enough to run on my laptop, handy when the wifi drops.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	LFM2.5 is small enough to run on my laptop, handy when the wifi drops.	null	null	null	null	null	small open-weights llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261101T093718Z_mw6z6o	null	-1	-69.6	-9.8	59.7	-56.9	-58.3	false	null	null	null	null	null	null	null	null	null	null	p_a5558ddb42	[]	{"mic_link": null, "bt_codec": null, "activity": "seated", "pronunciation_variants": [], "ui_note": null}
20261101T095841Z_x5ysv1	Have you tried LFM2.5 for the classification job yet?	5.013	[ "ai-models" ]	[ "LFM2.5" ]	ai-models_03b7668d29	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T09:58:41.677Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Have you tried LFM2.5 for the classification job yet?	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Have you tried LFM2.5 for the classification job yet?	null	null	null	null	null	small open-weights llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261101T095628Z_5bgz2g	null	-1	-70.9	-11	59.9	-56.9	-58.7	false	null	null	null	null	null	null	null	null	null	null	p_6dd667bc8a	[]	{"mic_link": null, "bt_codec": null, "activity": "seated", "pronunciation_variants": [], "ui_note": null}
20261101T194456Z_33yk52	Hy4 is the one from Tencent that I'd pick for longer chats, though I've barely used it.	5.775	[ "ai-models" ]	[ "Hy4" ]	ai-models_04bce0a674	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-01T19:44:56.848Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Hy4 is the one from Tencent that I'd pick for longer chats, though I've barely used it.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Hy4 is the one from Tencent that I'd pick for longer chats, though I've barely used it.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261101T194346Z_38cbaq	null	-1.4	-52.1	-17.6	34.5	-41.4	-42.4	false	null	null	null	null	null	null	null	null	null	null	p_47a7e648b7	[]	{"mic_link": null, "bt_codec": null, "activity": "seated", "pronunciation_variants": [], "ui_note": null}
20261102T094313Z_zj3y4q	Heads up, Hy4 wrote that whole reply, so don't be surprised by the tone.	4.885	[ "ai-models" ]	[ "Hy4" ]	ai-models_6de42352f9	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-02T09:43:13.978Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Heads up, Hy4 wrote that whole reply, so don't be surprised by the tone.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Heads up, Hy4 wrote that whole reply, so don't be surprised by the tone.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261102T091242Z_yu056l	null	-6.2	-55.1	-19.7	35.4	-90.3	-42.6	false	null	null	null	null	null	null	null	null	null	null	p_84a8d9e75b	[]	{"mic_link": null, "bt_codec": null, "activity": "seated", "pronunciation_variants": [], "ui_note": null}
20261102T154026Z_59od4c	Hey, try Seedream V4.5 on that product photo, the colours came out much warmer than the original.	6.453	[ "ai-models" ]	[ "Seedream V4.5" ]	ai-models_fb95ee71e9	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-02T15:40:26.168Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Hey, try Seedream V4.5 on that product photo, the colours came out much warmer than the original.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Hey, try Seedream V4.5 on that product photo, the colours came out much warmer than the original.	null	null	null	null	null	image to image model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261102T153432Z_obz3yv	null	-1	-68.3	-11.1	57.2	-55.2	-55.7	false	null	null	null	null	null	null	null	null	null	null	p_0a803c3bd8	[]	{"mic_link": "bluetooth", "bt_codec": "unknown", "activity": "seated", "pronunciation_variants": [], "ui_note": null}
20261102T154236Z_oogt50	Honestly, Z Image Turbo is fine for rough drafts where speed matters more than detail.	5.466	[ "ai-models" ]	[ "Z Image Turbo" ]	ai-models_7157716fe4	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-02T15:42:36.051Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Honestly, Z Image Turbo is fine for rough drafts where speed matters more than detail.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Honestly, Z Image Turbo is fine for rough drafts where speed matters more than detail.	null	null	null	null	null	text to image model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261102T153432Z_obz3yv	null	-1	-68.8	-11	57.8	-57.6	-51.7	false	null	null	null	null	null	null	null	null	null	null	p_ed76ad7d48	[]	{"mic_link": "bluetooth", "bt_codec": "unknown", "activity": "seated", "pronunciation_variants": [], "ui_note": null}
20261102T161608Z_zmh5d9	The final banner was made with Flux 2 Pro, so tell me if the lighting looks off.	4.762	[ "ai-models" ]	[ "Flux 2 Pro" ]	ai-models_7e1ec99d16	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-02T16:16:08.106Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	The final banner was made with Flux 2 Pro, so tell me if the lighting looks off.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	The final banner was made with Flux 2 Pro, so tell me if the lighting looks off.	null	null	null	null	null	text to image model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261102T153432Z_obz3yv	null	-1	-69.7	-11.1	58.6	-57.6	-57.4	false	null	null	null	null	null	null	null	null	null	null	p_68a6245c1c	[]	{"mic_link": "bluetooth", "bt_codec": "unknown", "activity": "seated", "pronunciation_variants": [], "ui_note": null}
20261102T161722Z_bkx2b1	Flux 2 Max took noticeably longer on the packaging mock-up, but the text was sharper.	5.407	[ "ai-models" ]	[ "Flux 2 Max" ]	ai-models_714b87ddca	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-02T16:17:22.871Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Flux 2 Max took noticeably longer on the packaging mock-up, but the text was sharper.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Flux 2 Max took noticeably longer on the packaging mock-up, but the text was sharper.	null	null	null	null	null	image to image model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261102T153432Z_obz3yv	null	-1	-69.6	-10.6	59.1	-58.1	-57.2	false	null	null	null	null	null	null	null	null	null	null	p_5575ac9bf5	[]	{"mic_link": "bluetooth", "bt_codec": "unknown", "activity": "seated", "pronunciation_variants": [], "ui_note": null}
20261103T065204Z_yez9wh	For the screenshots this week, Nemotron 3.5 Super VL is the model I'd pick for reading charts.	6.917	[ "ai-models" ]	[ "Nemotron 3.5 Super VL" ]	ai-models_cba61f9ef1	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-03T06:52:04.095Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	For the screenshots this week, Nemotron 3.5 Super VL is the model I'd pick for reading charts.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	For the screenshots this week, Nemotron 3.5 Super VL is the model I'd pick for reading charts.	null	null	null	null	null	vision llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261103T065031Z_5dyhy2	null	-1	-69	-10.6	58.4	-41.1	-52.8	false	null	null	null	null	null	null	null	null	null	null	p_383e85afd2	[]	{"mic_link": "bluetooth", "bt_codec": "unknown", "activity": "seated", "pronunciation_variants": [], "ui_note": null}
20261103T065536Z_2n0u0m	Seedance V1.5 Pro turned the still of the harbour into a short clip with decent motion.	6.079	[ "ai-models" ]	[ "Seedance V1.5 Pro" ]	ai-models_323540da51	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-03T06:55:36.153Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Seedance V1.5 Pro turned the still of the harbour into a short clip with decent motion.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Seedance V1.5 Pro turned the still of the harbour into a short clip with decent motion.	null	null	null	null	null	image to video model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261103T065031Z_5dyhy2	null	-1	-69	-10.4	58.7	-45.9	-56.2	false	null	null	null	null	null	null	null	null	null	null	p_f328537c63	[]	{"mic_link": "bluetooth", "bt_codec": "unknown", "activity": "seated", "pronunciation_variants": [], "ui_note": null}
20261103T065917Z_gtwxgz	Trying Gemini 4 Argon on the planning notes tonight, will report back on how it reads.	5.722	[ "ai-models" ]	[ "Gemini 4 Argon" ]	ai-models_843f2275d1	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-03T06:59:17.817Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Trying Gemini 4 Argon on the planning notes tonight, will report back on how it reads.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Trying Gemini 4 Argon on the planning notes tonight, will report back on how it reads.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261103T065031Z_5dyhy2	null	-1	-69	-11.5	57.6	-58.5	-56	false	null	null	null	null	null	null	null	null	null	null	p_f38befea14	[]	{"mic_link": "bluetooth", "bt_codec": "unknown", "activity": "seated", "pronunciation_variants": [], "ui_note": null}
20261103T065943Z_j0jdbf	Seedance 2.0 is the one I'd try first for the product spin, the motion looks smoother.	5.669	[ "ai-models" ]	[ "Seedance 2.0" ]	ai-models_a4519d6515	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-03T06:59:43.138Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Seedance 2.0 is the one I'd try first for the product spin, the motion looks smoother.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Seedance 2.0 is the one I'd try first for the product spin, the motion looks smoother.	null	null	null	null	null	image to video model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261103T065031Z_5dyhy2	null	-1	-70.2	-10.4	59.8	-53.8	-48.1	false	null	null	null	null	null	null	null	null	null	null	p_af7ee80daa	[]	{"mic_link": "bluetooth", "bt_codec": "unknown", "activity": "seated", "pronunciation_variants": [], "ui_note": null}
20261103T070041Z_vbppjo	I used Gemini 3.1 Flash TTS for the voice-over, and the pacing felt natural.	6.255	[ "ai-models" ]	[ "Gemini 3.1 Flash TTS" ]	ai-models_d95dde83e9	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-03T07:00:41.938Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	I used Gemini 3.1 Flash TTS for the voice-over, and the pacing felt natural.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	I used Gemini 3.1 Flash TTS for the voice-over, and the pacing felt natural.	null	null	null	null	null	text to speech model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261103T065031Z_5dyhy2	null	-1	-69.2	-11	58.2	-56.9	-56.7	false	null	null	null	null	null	null	null	null	null	null	p_81ccef6919	[]	{"mic_link": "bluetooth", "bt_codec": "unknown", "activity": "seated", "pronunciation_variants": [], "ui_note": null}
20261103T070255Z_i5jc8t	Should I pick MiniMax H3 for a short walking clip or stick with the old render?	5.429	[ "ai-models" ]	[ "MiniMax H3" ]	ai-models_7775d2b1d4	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-03T07:02:55.339Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Should I pick MiniMax H3 for a short walking clip or stick with the old render?	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Should I pick MiniMax H3 for a short walking clip or stick with the old render?	null	null	null	null	null	image to video model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261103T065031Z_5dyhy2	null	-1	-69.5	-10.6	58.9	-56.7	-22.6	false	null	null	null	null	null	null	null	null	null	null	p_92988b04da	[]	{"mic_link": "bluetooth", "bt_codec": "unknown", "activity": "seated", "pronunciation_variants": [], "ui_note": null}
20261103T070350Z_gqaq4b	I switched the clip to Kling v3 because the motion on people's hands looked far more natural.	6.199	[ "ai-models" ]	[ "Kling v3" ]	ai-models_f270d1e4b5	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-03T07:03:50.130Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	I switched the clip to Kling v3 because the motion on people's hands looked far more natural.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	I switched the clip to Kling v3 because the motion on people's hands looked far more natural.	null	null	null	null	null	image to video model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261103T065031Z_5dyhy2	null	-1	-68.3	-10.5	57.8	-42.8	-42.1	false	null	null	null	null	null	null	null	null	null	null	p_dadac6e5be	[]	{"mic_link": "bluetooth", "bt_codec": "unknown", "activity": "seated", "pronunciation_variants": [], "ui_note": null}
20261103T071629Z_7k9ihy	Gemini 3.8 Flash TTS did the voice-over for the onboarding video, and the pacing sounds right.	6.589	[ "ai-models" ]	[ "Gemini 3.8 Flash TTS" ]	ai-models_5abc8ab188	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-03T07:16:29.153Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	Gemini 3.8 Flash TTS did the voice-over for the onboarding video, and the pacing sounds right.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Gemini 3.8 Flash TTS did the voice-over for the onboarding video, and the pacing sounds right.	null	null	null	null	null	text to speech model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261103T065031Z_5dyhy2	null	-1	-70.1	-10.9	59.2	-58.9	-58.1	false	null	null	null	null	null	null	null	null	null	null	p_0e588cf223	[]	{"mic_link": "bluetooth", "bt_codec": "unknown", "activity": "seated", "pronunciation_variants": [], "ui_note": null}
20261103T071736Z_ckgsjx	The draft email was written by GLM 5.3, so please check the tone before it goes out.	5.978	[ "ai-models" ]	[ "GLM 5.3" ]	ai-models_78639eb54d	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	outdoor	48,000	2026-11-03T07:17:36.489Z	Yealink BH71	dataset	null	[]	null	8992	[]	null	null	null	The draft email was written by GLM 5.3, so please check the tone before it goes out.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	The draft email was written by GLM 5.3, so please check the tone before it goes out.	null	null	null	null	null	open-weights llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261103T065031Z_5dyhy2	null	-1	-69.6	-11.7	57.9	-58.1	-56.9	false	null	null	null	null	null	null	null	null	null	null	p_5f5515bd8c	[]	{"mic_link": "bluetooth", "bt_codec": "unknown", "activity": "seated", "pronunciation_variants": [], "ui_note": null}
20261105T132341Z_wd4vkq	Can you check which ARTPARK checkpoint we downloaded, the one from Hugging Face or the other?	5.215	[ "ai-models" ]	[ "ARTPARK" ]	ai-models_f10026544c	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-05T13:23:41.573Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Can you check which ARTPARK checkpoint we downloaded, the one from Hugging Face or the other?	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Can you check which ARTPARK checkpoint we downloaded, the one from Hugging Face or the other?	null	null	null	null	null	AI lab	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261105T132233Z_2bmdfu	null	-1.8	-46.8	-17.6	29.2	-36.5	-34.3	false	null	null	null	null	null	null	null	null	null	null	p_102d669869	[]	{"mic_link": "internal", "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261105T134202Z_n22rcv	Planning the week, Claude Opus 5.5 is for the long reports, not the quick replies.	5.669	[ "ai-models" ]	[ "Claude Opus 5.5" ]	ai-models_280b6c8e8f	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-05T13:42:02.346Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Planning the week, Claude Opus 5.5 is for the long reports, not the quick replies.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Planning the week, Claude Opus 5.5 is for the long reports, not the quick replies.	null	null	null	null	null	llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261105T132938Z_pjhyj5	null	-3.2	-43.8	-15.8	28.1	-34.3	-31.2	false	null	null	null	null	null	null	null	null	null	null	p_9b86db837f	[]	{"mic_link": "internal", "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261105T134929Z_93pdvx	Can you check whether Kimi K2.7 Code handles the refactor properly?	5.093	[ "ai-models" ]	[ "Kimi K2.7 Code" ]	ai-models_20ad67651d	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-05T13:49:29.167Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Can you check whether Kimi K2.7 Code handles the refactor properly?	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Can you check whether Kimi K2.7 Code handles the refactor properly?	null	null	null	null	null	coding model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261105T134555Z_zki1ad	null	-1.4	-37.3	-14.2	23.1	-29.2	-27.4	false	null	null	null	null	null	null	null	null	null	null	p_6fb7b89ecb	[]	{"mic_link": "internal", "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261105T135033Z_ba7wq6	I made three cover options with Ideogram 2a, so pick whichever looks least odd.	5.687	[ "ai-models" ]	[ "Ideogram 2a" ]	ai-models_c26ba834e9	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-05T13:50:33.386Z	internal	dataset	null	[]	null	8992	[]	null	null	I made 3 cover options with Ideogram 2a, so pick whichever looks least odd.	I made 3 cover options with Ideogram 2a, so pick whichever looks least odd.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	I made 3 cover options with Ideogram 2a, so pick whichever looks least odd.	null	null	null	null	null	text to image model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261105T134555Z_zki1ad	null	-1.7	-43.8	-14.8	28.9	-32.6	-33.8	false	null	null	null	null	null	null	null	null	null	null	p_2b46153b34	[]	{"mic_link": "internal", "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261105T135455Z_ap4fb6	Note to self: check whether Timestral needs the dates in daily steps before I rerun the inventory forecast.	6.426	[ "ai-models" ]	[ "Timestral" ]	ai-models_5178fad819	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-05T13:54:55.011Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Note to self: check whether Timestral needs the dates in daily steps before I rerun the inventory forecast.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Note to self: check whether Timestral needs the dates in daily steps before I rerun the inventory forecast.	null	null	null	null	null	time series forecasting model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261105T134555Z_zki1ad	null	-3.2	-46.8	-18.1	28.7	-37.9	-37.7	false	null	null	null	null	null	null	null	null	null	null	p_68c89dc766	[]	{"mic_link": "internal", "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261105T135755Z_ozvovq	I tried Nano Banana 2 on the Tuesday mock-ups and the lettering held up well.	4.911	[ "ai-models" ]	[ "Nano Banana 2" ]	ai-models_874bb781d2	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-05T13:57:55.929Z	internal	dataset	null	[]	null	8992	[]	null	null	null	I tried Nano Banana 2 on the Tuesday mock-ups and the lettering held up well.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	I tried Nano Banana 2 on the Tuesday mock-ups and the lettering held up well.	null	null	null	null	null	image model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261105T135516Z_fwqko2	null	-1.8	-38.1	-14.2	24	-28.5	-28.9	false	null	null	null	null	null	null	null	null	null	null	p_6c0e05b8d0	[]	{"mic_link": "internal", "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261106T083800Z_wscry0	Note to self: PixVerse V5.6 gave the cleanest text-to-video clip of the morning, so keep it.	9.301	[ "ai-models" ]	[ "PixVerse V5.6" ]	ai-models_a3bf57de2a	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-06T08:38:00.896Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Note to self: PixVerse V5.6 gave the cleanest text-to-video clip of the morning, so keep it.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Note to self: PixVerse V5.6 gave the cleanest text-to-video clip of the morning, so keep it.	null	null	null	null	null	text to video model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261106T081646Z_z4ahy7	null	-5.1	-46.4	-20.5	25.9	-23.9	-26.2	false	null	null	null	null	null	null	null	null	null	null	p_1958a9f68a	[]	{"mic_link": "internal", "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261106T084033Z_8keb3e	Mate, I'm using Lyria 3 Pro for the background track, so ping me if it sounds off.	7.386	[ "ai-models" ]	[ "Lyria 3 Pro" ]	ai-models_a1e5a0b011	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-06T08:40:33.815Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Mate, I'm using Lyria 3 Pro for the background track, so ping me if it sounds off.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Mate, I'm using Lyria 3 Pro for the background track, so ping me if it sounds off.	null	null	null	null	null	music model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261106T081646Z_z4ahy7	null	-2.1	-47.5	-19.8	27.7	-35.5	-34.9	false	null	null	null	null	null	null	null	null	null	null	p_895d51208b	[]	{"mic_link": "internal", "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261106T093422Z_3attf7	Can you run Pixai Tagger v1.0 over the whole folder before we pick the thumbnails?	6.122	[ "ai-models" ]	[ "Pixai Tagger v1.0" ]	ai-models_40e7ede5c9	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-06T09:34:22.203Z	Jabra EVOLVE 30 II	dataset	null	[]	null	8992	[]	null	null	null	Can you run Pixai Tagger v1.0 over the whole folder before we pick the thumbnails?	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Can you run Pixai Tagger v1.0 over the whole folder before we pick the thumbnails?	null	null	null	null	null	image classification model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261106T090608Z_0v39o4	null	-1	-46.3	-9.7	36.6	-27.2	-31.7	false	null	null	null	null	null	null	null	null	null	null	p_20e3f01345	[]	{"mic_link": "wired", "bt_codec": null, "mic_name": "Jabra EVOLVE 30 II", "mic_usb_vendor_id": "0b0e", "mic_usb_product_id": "0314", "activity": "seated", "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261106T094456Z_t7oxjj	Kling 4K takes a long time per run, so I only use it for the final cut.	4.762	[ "ai-models" ]	[ "Kling 4K" ]	ai-models_cf2c02a4e4	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-06T09:44:56.199Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Kling 4K takes a long time per run, so I only use it for the final cut.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Kling 4K takes a long time per run, so I only use it for the final cut.	null	null	null	null	null	video to video model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261106T090608Z_0v39o4	null	-27.5	-73.4	-68.7	4.8	-51.7	-45.3	false	null	null	null	null	null	null	null	null	null	null	p_ef24545628	[]	{"mic_link": null, "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261106T100147Z_wkr9qs	Pointed Devstral 2 at the broken test suite this afternoon and it found the bug quickly.	5.317	[ "ai-models" ]	[ "Devstral 2" ]	ai-models_17f84b5525	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-06T10:01:47.393Z	Jabra EVOLVE 30 II	dataset	null	[]	null	8992	[]	null	null	null	Pointed Devstral 2 at the broken test suite this afternoon and it found the bug quickly.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Pointed Devstral 2 at the broken test suite this afternoon and it found the bug quickly.	null	null	null	null	null	coding model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261106T095956Z_qw8fhp	null	-1	-57.4	-13.1	44.3	-20	-5.6	false	null	null	null	null	null	null	null	null	null	null	p_66f7e6a63c	[]	{"mic_link": "wired", "bt_codec": null, "mic_name": "Jabra EVOLVE 30 II", "mic_usb_vendor_id": "0b0e", "mic_usb_product_id": "0314", "activity": "seated", "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261106T100208Z_x7qdyo	Let's switch the narration over to Supertonic 3 before Thursday, the current clips sound flat.	6.383	[ "ai-models" ]	[ "Supertonic 3" ]	ai-models_9490a2a2ad	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-06T10:02:08.850Z	Jabra EVOLVE 30 II	dataset	null	[]	null	8992	[]	null	null	null	Let's switch the narration over to Supertonic 3 before Thursday, the current clips sound flat.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Let's switch the narration over to Supertonic 3 before Thursday, the current clips sound flat.	null	null	null	null	null	text to speech model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261106T095956Z_qw8fhp	null	-1	-60.1	-12.4	47.6	-49.3	-50.6	false	null	null	null	null	null	null	null	null	null	null	p_bb4247991f	[]	{"mic_link": "wired", "bt_codec": null, "mic_name": "Jabra EVOLVE 30 II", "mic_usb_vendor_id": "0b0e", "mic_usb_product_id": "0314", "activity": "seated", "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261106T103110Z_rihrgt	Render time on Fibo 2.0 Dev feels fine for thumbnails, though I haven't timed it properly.	6.511	[ "ai-models" ]	[ "Fibo 2.0 Dev" ]	ai-models_0ac4187f56	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-06T10:31:10.517Z	Jabra EVOLVE 30 II	dataset	null	[]	null	8992	[]	null	null	null	Render time on Fibo 2.0 Dev feels fine for thumbnails, though I haven't timed it properly.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Render time on Fibo 2.0 Dev feels fine for thumbnails, though I haven't timed it properly.	null	null	null	null	null	text to image model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261106T095956Z_qw8fhp	null	-1	-62.7	-11.3	51.4	-18.8	-44.5	false	null	null	null	null	null	null	null	null	null	null	p_76746ca891	[]	{"mic_link": "wired", "bt_codec": null, "mic_name": "Jabra EVOLVE 30 II", "mic_usb_vendor_id": "0b0e", "mic_usb_product_id": "0314", "activity": "seated", "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261106T103352Z_byov1o	Speech 2.8 HD sounds more natural than the older voice, so I'll use it for the audiobook intro.	6.365	[ "ai-models" ]	[ "Speech 2.8 HD" ]	ai-models_851d0ae776	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-06T10:33:52.858Z	Jabra EVOLVE 30 II	dataset	null	[]	null	8992	[]	null	null	null	Speech 2.8 HD sounds more natural than the older voice, so I'll use it for the audiobook intro.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Speech 2.8 HD sounds more natural than the older voice, so I'll use it for the audiobook intro.	null	null	null	null	null	text to speech model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261106T095956Z_qw8fhp	null	-1	-62.5	-11.9	50.7	-41.3	-48.2	false	null	null	null	null	null	null	null	null	null	null	p_968c846dab	[]	{"mic_link": "wired", "bt_codec": null, "mic_name": "Jabra EVOLVE 30 II", "mic_usb_vendor_id": "0b0e", "mic_usb_product_id": "0314", "activity": "seated", "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261106T105333Z_3z8hfg	Krea 2 Medium Turbo gave me 12 variations of the logo mark before the coffee went cold.	6.122	[ "ai-models" ]	[ "Krea 2 Medium Turbo" ]	ai-models_3b537c150b	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-06T10:53:33.860Z	Yealink BH70 Headset	dataset	null	[]	null	8992	[]	null	null	null	Krea 2 Medium Turbo gave me 12 variations of the logo mark before the coffee went cold.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Krea 2 Medium Turbo gave me 12 variations of the logo mark before the coffee went cold.	null	null	null	null	null	text to image model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261106T095956Z_qw8fhp	null	-1	-80.7	-9.8	70.9	-120	-69.5	false	null	null	null	null	null	null	null	null	null	null	p_0540b8da71	[]	{"mic_link": "wired", "bt_codec": null, "mic_name": "Yealink BH70 Headset", "mic_usb_vendor_id": "6993", "mic_usb_product_id": "b21a", "activity": "seated", "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261107T122440Z_mhud91	Tip for anyone dictating in Darija: Moulsot v0.3 copes with the code switching better than I expected.	8.197	[ "ai-models" ]	[ "Moulsot v0.3" ]	ai-models_85ddf6396e	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-07T12:24:40.309Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Tip for anyone dictating in Darija: Moulsot v0.3 copes with the code switching better than I expected.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Tip for anyone dictating in Darija: Moulsot v0.3 copes with the code switching better than I expected.	null	null	null	null	null	automatic speech recognition model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261107T121817Z_ktuu28	null	-3.9	-52.4	-16	36.4	-38.9	-35.7	false	null	null	null	null	null	null	null	null	null	null	p_54c9673de0	[]	{"mic_link": null, "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "background_speech_kinds": [], "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261107T163812Z_ouucvw	Email from Dara: the photo labels came back from Moondream 3, please have a look.	5.354	[ "ai-models" ]	[ "Moondream 3" ]	ai-models_63b9dba408	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-07T16:38:12.678Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Email from Dara: the photo labels came back from Moondream 3, please have a look.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Email from Dara: the photo labels came back from Moondream 3, please have a look.	null	null	null	null	null	small vision model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261107T163155Z_fejn9a	null	-3	-50.9	-13.5	37.4	-40.4	-41.2	false	null	null	null	null	null	null	null	null	null	null	p_32a54872c0	[]	{"mic_link": null, "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "background_speech_kinds": [], "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261107T164452Z_4dlm5a	Honestly, Flux 3 Image Klein is my go-to on Replicate when a client wants the same scene again.	6.453	[ "ai-models" ]	[ "Flux 3 Image Klein" ]	ai-models_8177f2161a	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-07T16:44:52.319Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Honestly, Flux 3 Image Klein is my go-to on Replicate when a client wants the same scene again.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Honestly, Flux 3 Image Klein is my go-to on Replicate when a client wants the same scene again.	null	null	null	null	null	image to image model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261107T163155Z_fejn9a	null	-3.6	-50.4	-16.3	34.1	-37.3	-41.4	false	null	null	null	null	null	null	null	null	null	null	p_463b49b761	[]	{"mic_link": null, "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "background_speech_kinds": [], "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261107T164528Z_fh82yd	Thinking of switching our drafting script over to Nanbeige4.2 for the cheaper runs on Monday.	6.202	[ "ai-models" ]	[ "Nanbeige4.2" ]	ai-models_07a72af54f	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-07T16:45:28.233Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Thinking of switching our drafting script over to Nanbeige4.2 for the cheaper runs on Monday.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Thinking of switching our drafting script over to Nanbeige4.2 for the cheaper runs on Monday.	null	null	null	null	null	text generation model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261107T163155Z_fejn9a	null	-2.6	-49.6	-14.9	34.7	-39.7	-39.4	false	null	null	null	null	null	null	null	null	null	null	p_a58e249049	[]	{"mic_link": null, "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "background_speech_kinds": [], "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261107T164535Z_0r0h5h	When the screenshot comes in, I ask SenseNova Loop what the error message says and what to fix.	6.607	[ "ai-models" ]	[ "SenseNova Loop" ]	ai-models_344e3dbd4b	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-07T16:45:35.072Z	internal	dataset	null	[]	null	8992	[]	null	null	null	When the screenshot comes in, I ask SenseNova Loop what the error message says and what to fix.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	When the screenshot comes in, I ask SenseNova Loop what the error message says and what to fix.	null	null	null	null	null	image text to text model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261107T163155Z_fejn9a	null	-3.1	-50.5	-14.7	35.9	-41.5	-39.6	false	null	null	null	null	null	null	null	null	null	null	p_2243945f6f	[]	{"mic_link": null, "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "background_speech_kinds": [], "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261107T165230Z_g3uo57	Could you try Mureka 8 on the rough vocal stem before we pick a final take?	5.306	[ "ai-models" ]	[ "Mureka 8" ]	ai-models_d1658a6cf0	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-07T16:52:30.738Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Could you try Mureka 8 on the rough vocal stem before we pick a final take?	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Could you try Mureka 8 on the rough vocal stem before we pick a final take?	null	null	null	null	null	audio to audio model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261107T163155Z_fejn9a	null	-2.3	-50.9	-14.2	36.6	-40.5	-40	false	null	null	null	null	null	null	null	null	null	null	p_c94410918e	[]	{"mic_link": null, "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "background_speech_kinds": [], "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261107T165430Z_imbzqd	Trying Nucleus Image tonight for the illustrations, will message you the results tomorrow.	5.231	[ "ai-models" ]	[ "Nucleus Image" ]	ai-models_85415302b6	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-07T16:54:30.787Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Trying Nucleus Image tonight for the illustrations, will message you the results tomorrow.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Trying Nucleus Image tonight for the illustrations, will message you the results tomorrow.	null	null	null	null	null	text to image model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261107T163155Z_fejn9a	null	-3	-49.6	-16.2	33.4	-39.1	-39	false	null	null	null	null	null	null	null	null	null	null	p_7257393882	[]	{"mic_link": null, "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "background_speech_kinds": [], "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261107T165931Z_ldvuz2	Dear team, the summary was made with DeepSeek V4 Flash, so flag anything that looks off.	6.213	[ "ai-models" ]	[ "DeepSeek V4 Flash" ]	ai-models_abfab0928c	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-07T16:59:31.521Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Dear team, the summary was made with DeepSeek V4 Flash, so flag anything that looks off.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Dear team, the summary was made with DeepSeek V4 Flash, so flag anything that looks off.	null	null	null	null	null	open-weights llm	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261107T163155Z_fejn9a	null	-3.6	-50.4	-15.1	35.3	-41	-40	false	null	null	null	null	null	null	null	null	null	null	p_e692c027c1	[]	{"mic_link": null, "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "background_speech_kinds": [], "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261107T215241Z_uhb7zo	Could you change the background in the client shot with Flux 3, keeping the lighting?	6.274	[ "ai-models" ]	[ "Flux 3" ]	ai-models_c0ccaf9c7e	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-07T21:52:41.518Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Could you change the background in the client shot with Flux 3, keeping the lighting?	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Could you change the background in the client shot with Flux 3, keeping the lighting?	null	null	null	null	null	image to image model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261107T213449Z_kylmnu	null	-1.3	-36.8	-15.8	21	-18.8	-27.7	false	null	null	null	null	null	null	null	null	null	null	p_69203c4a0e	[]	{"mic_link": null, "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "background_speech_kinds": [], "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261107T215355Z_bkvsdi	Use Hunyuan 3D 3.1 for the chair model, it gave me the cleanest mesh last night.	8.338	[ "ai-models" ]	[ "Hunyuan 3D 3.1" ]	ai-models_95fa30f8dc	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-07T21:53:55.934Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Use Hunyuan 3D 3.1 for the chair model, it gave me the cleanest mesh last night.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Use Hunyuan 3D 3.1 for the chair model, it gave me the cleanest mesh last night.	null	null	null	null	null	3d models model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	false	null	script	null	null	null	null	null	20261107T213449Z_kylmnu	null	-1.2	-28.3	-12.1	16.2	-19	-11.2	false	null	null	null	null	null	null	null	null	null	null	p_7f517af251	[]	{"mic_link": null, "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "background_speech_kinds": [], "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261107T221505Z_56urrc	Note to self: the HeyGen Voice Turbo output sounds a bit flat on questions, so rewrite those lines.	6.087	[ "ai-models" ]	[ "HeyGen Voice Turbo" ]	ai-models_ff7257b302	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-07T22:15:05.792Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Note to self: the HeyGen Voice Turbo output sounds a bit flat on questions, so rewrite those lines.	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Note to self: the HeyGen Voice Turbo output sounds a bit flat on questions, so rewrite those lines.	null	null	null	null	null	text to speech model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261107T213449Z_kylmnu	null	-4.7	-57.4	-19	38.4	-51.1	-48.4	false	null	null	null	null	null	null	null	null	null	null	p_6b95d80d2b	[]	{"mic_link": null, "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "background_speech_kinds": [], "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}
20261107T221554Z_oke1ag	Could you run the scanned invoices through MiniCPM OCR 1.0 and send me the totals?	7.098	[ "ai-models" ]	[ "MiniCPM OCR 1.0" ]	ai-models_dd71e77704	claude-haiku-5-5	bank	false	OnePlus Nord 3 5G	quiet-indoor	48,000	2026-11-07T22:15:54.537Z	internal	dataset	null	[]	null	8992	[]	null	null	null	Could you run the scanned invoices through MiniCPM OCR 1.0 and send me the totals?	null	[]	null	null	null	Daniel Rosehill	male	[]	false	ai-models	Could you run the scanned invoices through MiniCPM OCR 1.0 and send me the totals?	null	null	null	null	null	image text to text model	null	null	null	Daniel Rosehill	24	flac	48,000	stt	true	null	script	null	null	null	null	null	20261107T213449Z_kylmnu	null	-5	-48.6	-19.4	29.1	-40.9	-39.2	false	null	null	null	null	null	null	null	null	null	null	p_80df19cb8a	[]	{"mic_link": null, "bt_codec": null, "mic_name": "CPH2493", "mic_usb_vendor_id": null, "mic_usb_product_id": null, "activity": "seated", "background_speech_kinds": [], "pronunciation_variants": [], "ui_note": null, "source_url": null, "check_verdict": null, "check_errors": null, "check_error_types": null}

End of preview. Expand in Data Studio

AI Model Sentences

Audio dataset, split-out. Rows copied from the master, danielrosehill/Read-Aloud-Voice-Sentences, so one task can be browsed alone. Training reads the master, not this copy. Every audio dataset as one tree, with hours: the audio index. How every dataset fits together: the dataset index.

English sentences naming one current AI model each (GPT-6.1 Terra, Claude Opus 5.5, Qwen3.8 Max, Kimi K3.1 Pro), read by one speaker, spelled the vendor's way. Names come from the OpenRouter model list, newest first, skipping stealth/codename and deprecated models, and are topped up as new models are released. target_terms lists the model name; prompt_kind is its kind (llm, coding model, image model). No Hebrew. Prompts written by Claude Haiku 5.5 and reviewed (2026-10-31).

Website: hebrish.app: get the Hebrish keyboard for Android and the desktop apps, see the results, and find every model, dataset and demo in one place.

Two tabs. sentences lists every ground-truth sentence with a takes count, so takes = 0 shows what has not been recorded. takes has one row per recording with its audio, joined on sentence_id; a sentence can have several takes. load_dataset("Hebrish/AI-Model-Sentences", "sentences") reads the text; the default config reads the takes.

Takes come from danielrosehill/Read-Aloud-Voice-Sentences, the full dataset, with the same ids and train/test split, by scripts/sync_subsets.py in the project repository. Do not edit this dataset directly. New takes append and only the sentence counts are rewritten.

Speaker: Daniel Rosehill, the dataset's author, who was born in Ireland and speaks English with an Irish accent, and who lives in Jerusalem. Every recording is his voice. Deliberate variation: he sometimes changes his speaking speed on purpose (fast and rushed, or slow) and sometimes says a word the way people commonly mispronounce it, to make the models robust to real speech. The transcript is always the sentence as written, so a take whose speed or pronunciation sounds off against its card is usually intentional, not a recording error.

All audio was recorded on a OnePlus Nord 3 5G phone using its internal microphone, with browser echo cancellation, noise suppression and auto gain turned off.

Layout

sentences/train.parquet  one row per sentence: `id`, `text`, `takes`, `prompt_ids`, `served` and the sentence columns
data/train-*.parquet   one row per take: `audio` (the FLAC or WAV file, embedded) and every column
meta/train.jsonl       index: the same rows without audio, plus `shard`, `audio_path` and the audio's byte range
meta/schema.json       column types
data/test-*.parquet, meta/test.jsonl   held-out set, never trained on

Audio is 48 kHz mono 24-bit FLAC for rows from 2026-10-22 and 16 kHz mono 16-bit WAV before.

The test split is term-disjoint: a held-out term never appears in train. load_dataset("Hebrish/AI-Model-Sentences") works directly.

Columns

A take's sentence_id is its row's id in sentences. A sentence with served: false is no longer offered by the app; it is kept because a take used it. The columns are the same as in the main dataset: text (ground truth, punctuated and cased like Parakeet v2 output), text_he (the same transcript with its Hebrew terms in Hebrew script, for Hebrew-capable models), categories, target_terms, prompt_source, prompt_origin (manual / dataset / generated), edited, duration_s, environment, conditions, device, speaker, speaker_gender, contributor, hebrew_pronunciation, variant_groups, variant_spelling and recorded_at. The main dataset's card describes them.

Licence

CC-BY-4.0, with sentence-text attribution per prompt_source as described in the main dataset's licence section.

Downloads last month: 389