Repository navigation
feat(transcription): transcribe with OpenRouter - #540
Conversation
gpt-4o-transcribe-diarize rejects requests with more than 1400 s of audio, but MP3 chunks were split by size alone. Any MP3 under about 120 kbps put more than 1400 s in a 20 MB chunk, so every chunk failed. For diarization, split MP3 at frame boundaries into the fewest equal parts that stay under both 20 MB and 1400 s. Whisper keeps its size-only chunks. Fixes #532
A chunk that OpenAI rejected with a status such as 400 or 401 was sent three times, although each send fails the same way. A long diarized episode went up three times before the error showed. Retry only the statuses openai-node retries: 408, 409, 429 and 5xx, plus network errors. Refs #532
gpt-4o-transcribe-diarize takes up to 0.58 s per second of audio, measured over 81 requests, so the 1,072 s parts a 54 min episode got would take about 10 min each, at the request timeout. Parts are now at most 600 s, which the model answers in under 6 min. Refs #532
Each test passes on the hand-written code. Together they catch wrong derivations that no existing test sees: the -1 SecretStorage ID suffix that released versions accept, the provider an export error names, the kind a settings key row adopts a secret for, and the key migration notices.
CREDENTIALS in src/types/Credentials.ts holds each kind's settings field, export key and label. The credential repository, settings decoder, settings transfer, settings tab and key migration notice loop over it. IPodNotesSettings takes its key fields from it, so a kind with no default in DEFAULT_SETTINGS fails typecheck. The plaintext key map from before 2.19.0 stays fixed.
The published transcript renderer is the reader of the format that stays when the diarization folder goes. The default template now lives only in DEFAULT_SETTINGS, so src/constants.ts imports no service code.
src/services/transcription/ turns an episode into a transcript with any provider. A provider is one file: its models, its request, and how to read its answers. OpenAI (gpt-transcribe) and Deepgram (nova-3) are the first two. The engine owns what providers share: - audioSource.ts decides whether a provider may be given the episode's address. Only an address that a saved public feed lists, that carries no credential, and that names a public host passes. All other audio is uploaded. - split.ts is audioChunker.ts with the caps as arguments and a start time on each part. - http.ts sends a request at most three times, and only after a 429, a 5xx, or a lost connection. Its failure sentences never include the provider's own text, which can quote a private address. - cues.ts turns words or text into the shapes of a published transcript. Nothing calls the engine yet. Lint rejects an import of the folder's files from outside it, except its index.
TranscriptionService keeps its queue, the published-transcript path, the progress notice, and note saving. It calls the engine for audio. - OpenAI transcribes with gpt-transcribe. whisper-1 and gpt-4o-transcribe-diarize, which OpenAI shuts down on 2027-02-26, are gone. A tooling test searches the built main.js for them. - With speaker labels on and OpenAI selected, the transcript is saved without labels. A notice that stays until clicked names the providers that label speakers. - Deepgram transcripts go through renderTranscriptBody, so each paragraph has a timestamp. Speakers are numbered in the order they first speak. - Deepgram gets the address of a public episode and downloads it itself. - A failure notice stays until clicked. Parts go out one at a time, and a 401, 402, 404, or 400 stops the job before the next part. Deletes src/services/diarization/, openaiTranscription.ts, the Whisper worker pool, and createBinaryChunkFiles. getEpisodeAudioBuffer no longer returns a basename, because uploads are named for their format only.
net.mjs answers OpenAI with { text } and Deepgram with the words of two
speakers, and logs the content type of each request. fixtures.mjs serves a
feed at priv3.supercast.com, an address that PodNotes keeps private. The
transcripts recipe covers OpenAI in parts, and Deepgram by address and by
upload.
The marketplace review test called TranscriptionService.waitForRetry, which the engine replaced. An aborted run now rejects through waitForLifecycle, so the test aborts a pending wait there and expects the same Error.
An aborted job no longer downloads the episode or decodes it into parts. Until now only its requests were refused. A 429 or a 5xx that asks to wait over a minute is sent once. A sooner send would only upload the audio to be refused again.
… feat/transcription-providers
A provider's name was written twice: once in its credential row and once in its provider file. The row now owns it, so a provider file holds only its models and its request. Each model gains the name the settings will show.
The provider was derived from the speaker settings: Deepgram with speaker labels on, OpenAI otherwise. It is now stored as transcript.provider, with transcript.model beside it, where "" follows the provider's recommended model. schemaVersion stays 2 and both fields are additions, so a device on 2.24.0 keeps loading the library and keeps the fields on its own saves. A library that has no transcript.provider gets the provider it ran with: Deepgram for speaker labels with Deepgram, with or without its key, and OpenAI otherwise. The migration reads only the stored document, adds no decode warning, and gives the same answer on every load until a save stores it. transcript.diarization.provider belongs to versions up to 2.24.0. This version reads it for that migration alone, never writes it, keeps it in data.json as found, and leaves it out of settings exports. Importing a 2.24.0 export that labels speakers with Deepgram selects Deepgram. A provider or model ID this version does not have is kept as stored and shown under that name. An unknown provider refuses to transcribe with a notice. An unknown model runs the recommended model and the notice says so. The Transcripts settings now read: Transcription provider, the chosen provider's API key, Model, Speaker labels, Speaker label format, then the path and template. Speaker labels is disabled for a model that cannot label speakers. The Diarization provider row is gone.
ElevenLabs is a credential row, one provider file and a registry line. Its key reference is the new root field elevenLabsSecretId, stored in SecretStorage as podnotes-elevenlabs-api-key, and a settings export carries the key as secrets.elevenLabs. The export envelope stays at version 2. A public episode goes to ElevenLabs by its address, so nothing is downloaded. Any other episode is uploaded in one request. Word times and speakers reach the note, and turning speaker labels off asks for none. ElevenLabs answers 401 for a used-up quota as for a rejected key, and 400 for a model the key may not use as for an address it could not download, so the provider reads detail.code. Only the last of these is followed by an upload. A request waits 10 minutes for its answer. A new install starts on ElevenLabs with speaker labels on. A library saved up to 2.24.0 that holds no OpenAI key and has speaker labels off moves to ElevenLabs too, because it never set transcription up. A library that never stored the speaker labels choice keeps them off. The notice for a model without speaker labels now names ElevenLabs first.
A journey now seeds its transcription provider with its key, because a new vault is on ElevenLabs. The OpenAI journeys choose OpenAI. A new journey runs the default: a public episode goes to ElevenLabs by its address, no audio is fetched, and the note has timestamps and speaker labels. The ElevenLabs stub answers with the words ElevenLabs returns and records what each request named. The settings tab test walks the new Transcripts rows, and the key recovery test links the ElevenLabs key, the one a new vault shows.
The fixture provider answers api.elevenlabs.io with ElevenLabs' words, two speakers with the second 12 s in, and logs the address a request named in place of a file. The transcript guide follows the new settings rows and adds the ElevenLabs runs and a 2.24.0 library check.
…choice The provider row decided for itself whether the stored provider is one this version has, and the Speaker label format row read the stored speaker setting. Both now ask transcriptionChoice, as the key rows, the model row and the Speaker labels row do. The engine no longer exports a type nothing imports.
The Transcripts tests looked each control up by a CSS selector inside its row and asserted on a hidden row's inline style. They now query by role and by the row's name, as a user or a screen reader finds a control, so a hidden row is one the query does not find.
…ings measure The transcription engine left main.js at 103,795 gzip bytes. The provider and model settings, their migration and the ElevenLabs provider add 866, which takes it to 104,661 and over the 104,000 budget. The budget rises by those 866 bytes to 104,866, so the 205 bytes of headroom stay as they were.
…m library 2.24.0 imports the diarization object whole and fills a missing provider with OpenAI, so omitting the field moved Deepgram libraries to OpenAI. Imports here keep the provider this library stored.
A library migrated with OpenAI speaker labels on could not turn them off, so every transcription repeated the notice.
… one The tests for a provider and a model a newer PodNotes stored used OpenRouter and Grok STT 2.0, which this version is about to run. They now use AssemblyAI and universal-3-6-pro.
OpenRouter is a credential row, one provider file and a registry line. Its key reference is the new root field openRouterSecretId, stored in SecretStorage as podnotes-openrouter-api-key, and a settings export carries the key as secrets.openRouter. It offers three models, each with the request measured to give word times and speakers: - Grok STT 2.0, recommended: base64 JSON with verbose_json and word and segment times, routed to xAI with provider.only, which takes diarize from provider.options.xai. A request routed elsewhere is answered with no speakers. - ElevenLabs Scribe v2: multipart with diarize beside the model, or JSON with input_audio.url for an episode a saved public feed lists. - MAI-Transcribe 2: multipart with diarize beside the model. With speaker labels off, no request carries diarize. Sounds Scribe v2 lists as audio_event items are left out. OpenRouter takes 25 MB a request and gives the model 60 s, so an upload goes in parts of at most 18 MB and 1,800 s. Parts of 24 MB were answered in 20 to 31 s. A 402 says the account is out of credit, a 401 that the key was rejected, and a 404 that the model is not offered. A 404 whose message names the account's data policy or zero data retention says the model is not offered under the account's OpenRouter privacy settings. Each ends the job at the first part, and none quotes OpenRouter.
…t each part A note saved with speaker labels asked for and none returned now says so: "Saved without speaker labels, because Grok STT 2.0 returned none." OpenRouter answers 200 with no speakers when a request reaches a route that ignores the speaker option, so this is checked for every provider. Labels from two parts already name different speakers. An episode sent in parts now also says that its labels restart in each part, so one person can have several labels, until labels are merged across parts.
…ll lists PodNotes reads OpenRouter's public catalog of transcription models, without the key, when the Model row draws and when an OpenRouter job starts. It keeps the read in memory for 24 hours. A read that fails is not kept, so the next one asks again, and the tested list is offered as it is. The catalog only takes models away. A model it does not list is not offered, and one whose expiration_date has come is skipped. A model with a later end date reads "Grok STT 2.0, ends <date>" in the Model row. A catalog that would leave no model is not followed. A chosen model the catalog no longer lists stays selected in the Model row as "<label>, no longer offered". A job runs the first offered model instead and says "OpenRouter no longer offers ElevenLabs Scribe v2, so PodNotes used Grok STT 2.0." The row redraws when a read lands while it is open. A provider gets its catalog through an optional catalog() beside failure(), so the registry owns the cache and the filter for any provider that has one.
A stub answers OpenRouter's transcriptions with two speakers, the second 12 s in, and its model catalog with the three models PodNotes offers. It records what each request named and a hash of the audio it carried, decoded from base64 for JSON. J-TR-09 plays a public episode on Scribe v2: one JSON request names the episode's address, and the episode's audio is never fetched. J-TR-10 plays it on the recommended Grok STT 2.0: one JSON request carries the audio as base64, the same bytes the fixture server holds. Both notes have timestamps and numbered speakers, and the catalog is read once, with no Authorization. A journey seeds any provider by its credential row, and a model with it.
…eads them The fixture provider answers openrouter.ai transcriptions with verbose_json words from two speakers, the second 12 s in, and the model catalog with the three models PodNotes offers. The log keeps openrouter.ai paths and reads the address a JSON request names in input_audio.url. The transcript guide adds the OpenRouter rows, its runs (Scribe v2 by address, Grok by base64, a long MP3 in parts) and the notices that now name OpenRouter.
…ame parts Summing frame durations drifts past the cap: 25,000 frames of 24 ms sum to 600.0000000000218 s, so the per-frame seconds check cut each part a frame early and a 49,999-frame episode went as 24,999 + 24,999 + 1 frames. Parts are already counted in frames, which keeps them under maxSeconds, so the check goes.
The planner fixed a frame count per part and checked the size cap separately. When a leading tag or a variable bitrate put an equal split over the cap, the size check cut a part short and the next kept the old frame count, leaving a one-frame part that cannot be decoded: 43,690 frames of a 320 kbps MP3 with a 512-byte tag went as 21,844 + 21,845 + 1. The planner now takes the fewest equal splits that fit both caps, adding a part when an equal split does not fit.
Equal splits put the odd frame first, so a 512-byte tag and 43,689 frames of a 320 kbps MP3 put the first of two parts 192 bytes over 20 MiB, and the planner sent three where two fit. Each cut now takes an equal share of the frames still to place and steps back while the part is over the size cap, so 21,844 + 21,845 goes in two.
Cuts only moved back from an even split, so where the bitrate rose a part was added although moving the cut forward fit: 21,847 low-bitrate frames then 21,845 at 320 kbps and an ID3v1 tag went as three parts, not two. The planner now finds the fewest parts by filling each as far as it fits, then places each cut as close to an equal share as the parts before and after it allow. A property test checks the count against a brute-force minimum on 300 random variable-bitrate streams.
Frame-like bytes inside a leading tag set the first frame's duration, and counting every part in that frame's length sent a 1,728 s episode of 72 ms frames up in one request. Parts now fit by the summed duration of their own frames, and each part's start is the sum of the frames before it.
Audio frames inside a tag's declared size were counted and could be cut through, sending an undecodable first part, and frame-like bytes in a trailing tag added a part.
…engine # Conflicts: # src/services/TranscriptionService.test.ts # src/services/TranscriptionService.ts # src/services/audioChunker.test.ts # src/services/audioChunker.ts # src/services/diarization/openaiProvider.ts # src/services/diarization/types.ts # src/services/openaiTranscription.test.ts # src/services/openaiTranscription.ts # tests/mocks/openaiFailures.ts
…h cap A file in a format other than MP3 under the size cap went whole whatever its length, so a long .m4a passed a model's maxSeconds. Its length now comes from its metadata through a media element, without decoding it, and audio over the cap is decoded into WAV parts that fit both caps. Audio of an unknown length still goes whole.
A public feed's enclosure with AWS or Google Cloud signing parameters passed the feed credential check, so Deepgram got the whole signed address. A query parameter whose name holds signature, credential or token, or Akamai's hdnts, now uploads the audio instead.
…nload A WebView may load no metadata for media it does not play, so the length probe could wait forever, and an unload could not stop it before a decode. The probe now reads the length as unknown after 5 s or when the job is aborted, and an aborted job decodes nothing.
…nscription-providers # Conflicts: # AGENTS.md # scripts/size-baseline.md # scripts/size.mjs # scripts/size.test.ts # src/persistence/settingsCodec.ts # src/settingsTransfer.ts # src/types/Credentials.ts # src/ui/settings/PodNotesSettingsTab.ts
…nscription-openrouter # Conflicts: # AGENTS.md # scripts/size-baseline.md # scripts/size.mjs # scripts/size.test.ts
|
Retargeted to
|
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes using high effort and found 1 potential issue.
❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.
Reviewed by Cursor Bugbot for commit f60f448. Configure here.
|
FAIL
|
…address An address goes in one request, so Scribe v2 on OpenRouter got a public episode of any length by address, past the 1,800 s its parts are capped at for OpenRouter's 60 s limit. An episode the feed says is longer than the model's cap is now uploaded in parts. One of unknown length still goes by address.
|
On the catalog finding in the f60f448 verdict: falling back to the fixed list when the catalog would leave no model is intended, and documented on
If OpenRouter's catalog ever drops every model PodNotes knows, the fix is a release with new model IDs. The weekly model review in #541 exists to catch that. |
|
PASS
|

Summary
PodNotes is moving transcription to current models from several providers, so that a provider is a credential row, one provider file and a registry line, and no user runs a retired model. This PR is step 4 of that plan: OpenRouter. It stacks on #539.
Transcription provideroffersOpenRouterbetween Deepgram and OpenAI. Choosing it shows theOpenRouter API keyrow and aModelrow withRecommended (Grok STT 2.0),Grok STT 2.0,ElevenLabs Scribe v2andMAI-Transcribe 2.provider.only, which takesdiarizefromprovider.options.xai. A request routed elsewhere answers 200 with no speakers. Scribe v2 and MAI-Transcribe 2 get multipart withdiarizebeside the model. Scribe v2 gets the address of an episode a saved public feed lists, so nothing is downloaded. With speaker labels off, no request carriesdiarize.OpenRouter transcribed this episode in 3 parts, and speaker labels restart in each, so one person can have several labels.Merging labels across parts is the next PR.Saved without speaker labels, because Grok STT 2.0 returned none.This holds for every provider.expiration_datehas come is skipped, and a later end date readsGrok STT 2.0, ends <date>. A chosen model the catalog drops stays selected as<label>, no longer offered, and the job runs the recommended model with a notice. A failed read is not kept and leaves the tested list.time=1400.112, nottime=1400.1119999994578.For review:
catalog()besidefailure(). The registry keeps the reads in a Svelte store, so the Model row redraws when a read lands while it is open, and tests reset it through the same handle.privacy, which ends the job likemodel-unavailable, because the user's action differs.withdrawnModelsits besideunknownModelon the choice, because each says something different: this version has no such model, or the provider's catalog dropped it.Not measured with a real key, for the lead's real-key check:
/data policy|zero data retention|\bzdr\b/ionerror.message. No probe saved the real body. The pattern follows OpenRouter's published "No endpoints found matching your data policy (Zero data retention)".format: "mp3", as measured on an MP3 episode. An M4A episode by address is unmeasured.verbose_jsonwithoutdiarize. Whether they still returnwordsis unmeasured.OpenRouter could not transcribe this audio.with no upload, because OpenRouter's answer for it is unknown.Not in this PR: merged speaker labels across parts, the weekly model review, and the docs, which describe the released version until the release.
Verification
Run on Linux with Node 22.23.4, on the final tree
c46f215a:npm run lint,npm run format:check,npm run typecheck,npm run buildandnpm run docs:buildpass.npm run test: 124 files, 2,878 tests pass.npm run test:tooling: 8 files, 48 tests pass.npm run test:e2eon Obsidian 1.14.4: 33 files, 257 tests pass, J-TR-09 and J-TR-10 new.PODNOTES_OBSIDIAN_APP=/opt/Obsidian/obsidian npm run test:e2e -- settings-tab journey-resilienceon Obsidian 1.13.7: 2 files, 23 tests pass.npm run size: 106,031 gzip bytes, 131 under the new 106,162 budget. The base measures 104,706 in a separate build of1eef3e6f. The budget rises by the 1,296 bytes OpenRouter measured ata844ca70, and the millisecond fix after it adds 29.npm run plant: 27 planted defects, each caught by the unit test named below, and 2 more caught bynpm run test:e2e.Obsidian 1.14.4 on Linux under Xvfb, in this worktree's isolated vault, with the verify skill's fixture provider answering OpenRouter and its catalog. No real key was used. The fixture feed is
http://podnotes.test:24280/feed.xml. The local file isFixtures/Long Fixture.mp3, 4,200 s of sine at 8 kbps mono from ffmpeg, 4,200,452 bytes.data.jsonheldprovider: "openrouter"andmodel: "". The network log showed oneGET https://openrouter.ai/api/v1/modelswith no upload. The warning beside the key is Obsidian's ownSecrets are not encrypted, because Xvfb has no keyring.ElevenLabs Scribe v2chosen by keyboard,Transcribe current episode from audioonFixture Episode 1sent onePOST https://openrouter.ai/api/v1/audio/transcriptions,application/json, 160 bytes, naminghttp://podnotes.test:24280/episode-1.wav. The fixture server saw no request during the command. The note has**1:**at 0 s and**2:**at 12 s.Recommended (Grok STT 2.0), after a plugin reload, the command sent the catalogGET, then threePOSTs of 1,867,279, 1,867,027 and 1,866,931 bytes of JSON, one after the other. The note numbers six speakers attime=0,12,1400.112,1412.112,2800.224and2812.224, and the restart notice stayed until clicked. Before the millisecond fix the same run linkedtime=1400.1119999994578.dev:errorsreported no error. The instance and the fixture server are stopped. The lead verifies with a real key separately.Planted defects, each caught by the named test through
npm run plantprovider.onlyproviders/openrouter.tsproviders/openrouter.tsproviders/openrouter.tsproviders/openrouter.tsproviders/openrouter.tsdiarizedifferentlyproviders/openrouter.tsproviders/openrouter.tscues.tsproviders/openrouter.tscues.tscues.tstranscribe.tstranscribe.tshttp.tsproviders/openrouter.tstranscribe.tsregistry.tsTranscriptionService.tsregistry.tsregistry.tsregistry.tsregistry.tsexpiration_datenot readproviders/openrouter.tsPodNotesSettingsTab.tsPodNotesSettingsTab.tsPodNotesSettingsTab.tsproviders/openrouter.tsproviders/openrouter.tsnpm run test:e2e)providers/openrouter.tsnpm run test:e2e)Real OpenRouter key (lead, after review)
Live probes first, then Obsidian 1.14.4, Linux, build of
23a4d86b, palette command, a pass-through request log (no fixture):https://feeds.npr.org/500005/podcast.xml), labels on: one 520-byte POST naming the address, no download, saved in 5 s with timestamps and speakers. Labels off: one 505-byte POST; the note has timestamps and no speakers.dev:errors: none.Review fixes (lead)
verbose_jsonwith nowordswhen neitherdiarizenor word timestamps are asked for, so a labels-off job failed with "OpenRouter found no speech". Both now always send word timestamps (timestamp_granularities[]=word, or["word"]in JSON). Plant: drop it (caught bywith speaker labels off, elevenlabs/scribe-v2 is still asked for word times, which OpenRouter sends only when asked).unfetched, so the job uploads once. Plant: return no failure kind (caught byScribe v2 uploads the episode once when OpenRouter's provider cannot download its address)./data policy|zero data retention|\bzdr\b/i) matches the real ZDR body captured earlier: "0 endpoints out of 1 requested are available matching your guardrail restrictions and data policy ... ZDR violation (account settings)".Release impact
openRouterSecretId, empty by default. The key lives in SecretStorage aspodnotes-openrouter-api-key. A settings export carries it assecrets.openRouterwhen keys are included. The schema stays at version 2. feat(transcription): choose a provider and model, with ElevenLabs as the default #539 measured that 2.24.0 keepselevenLabsSecretId, an unknown root field to it. This field takes the same path, which I inferred and did not run.transcript.providercan now beopenrouter. A version without OpenRouter keeps the value and refuses to transcribe with a notice, as feat(transcription): choose a provider and model, with ElevenLabs as the default #539 does for any unknown provider.GETtoopenrouter.ai/api/v1/models?output_modalities=transcriptionwhen the Model row draws or a job starts. After a read that works it waits a day. After one that fails it asks again next time.Note
Add OpenRouter as a transcription provider with model catalog and multipart upload
modelRownow refreshes models on draw and can follow a supplied store viafollow(store);transcribeAudiorefreshes the provider catalog before resolving credentials; gzip budget rises to 106,562 bytes in size.mjs.Macroscope summarized df7e41c.
Note
Medium Risk
Touches external API calls, API key storage, and transcription job logic with catalog fallbacks; well-covered by unit and e2e tests but real-key edge cases remain partially unverified per PR notes.
Overview
Adds OpenRouter as a transcription provider with
openRouterSecretIdcredentials, settings (API key + model picker), and verify/e2e fixture support.The new provider supports Grok STT 2.0 (base64 JSON routed to xAI), ElevenLabs Scribe v2 (public episode URL when under 1,800s), and MAI-Transcribe 2, with 18 MB / 1,800 s chunking, per-part speaker numbering, and user-facing warnings for missing labels, withdrawn models, and multi-part restarts.
A daily-cached public model catalog filters offered models, marks end dates in the UI, falls back when a stored model is delisted, and
refreshModelsruns before jobs and when the model row draws. Adds aprivacyfailure kind for OpenRouter ZDR/policy 404s.Transcript cue times are rounded to milliseconds for seek links. The gzip bundle budget rises to 106,562 bytes.
Reviewed by Cursor Bugbot for commit df7e41c. Bugbot is set up for automated code reviews on this repo. Configure here.