Speech Transcribe
Use the audio_transcribe tool when the user asks to transcribe,
caption, diarize, timestamp, or identify speakers in an audio or video clip.
The tool comes from the media-tools plugin. If it is not in your tool list,
tell the user to enable it with hybridclaw plugin install media-tools.
Workflow
- Call
audio_transcribe with action: "list" if you need provider
readiness or the user asks what is configured.
- Pass
audio as a current attachment filename/ref, /workspace path,
/discord-media-cache path, /uploaded-media-cache path, or HTTPS media
URL.
- Use
provider: "auto" unless the user asks for openai, deepgram, or
assemblyai, or unless diarization is required. Auto mode honors the tenant
skills.speechToText.defaultProvider config when it is set. Prefer Deepgram
or AssemblyAI for speaker labels.
- Pass
language only when the user gives a known language. Omit it for
provider language detection.
- Set
diarization: true when the user asks for speaker labels.
- Set
timestamps to word, segment, or none based on the request.
- Return the structured result fields that matter: transcript text, provider,
detected language, duration, cost, warnings, and artifact paths.
The tool owns provider credentials, provider fallback, output schema,
long-audio chunking for local and remote OpenAI uploads when ffmpeg/ffprobe
are available, transcript artifact persistence, and usage-cost accounting.
Output Contract
The tool returns JSON shaped like:
{
"text": "Transcript text",
"segments": [{ "start": 0, "end": 1.2, "speaker": "speaker_0", "text": "..." }],
"language": "en",
"provider": "deepgram",
"duration_sec": 12.3,
"cost_usd": 0.001
}
Transcript text and segment JSON are also persisted as private workspace
artifacts. Treat transcripts as operator-private until the user explicitly asks
to share, post, email, or publish them.
1---2name: speech-transcribe3description: Transcribe audio with the audio_transcribe tool, including provider override, diarization, timestamps, language detection, and transcript artifacts.4user-invocable: true5requires:6 bins:7 - node8credentials:9 - id: openai-api-key10 kind: api_key11 required: false12 secret_ref:13 source: store14 id: OPENAI_API_KEY15 scope: OpenAI Whisper speech-to-text transcription.16 how_to_obtain: "Create an OpenAI API key. Set `OPENAI_API_KEY` through browser admin at the active `/admin/secrets` route; if browser admin is unavailable, use `/secret set OPENAI_API_KEY \"<api-key>\"` in browser `/chat` or TUI; local console fallback: `hybridclaw secret set OPENAI_API_KEY \"<api-key>\"`."17 - id: deepgram-api-key18 kind: api_key19 required: false20 secret_ref:21 source: store22 id: DEEPGRAM_API_KEY23 scope: Deepgram speech-to-text transcription, diarization, and word timestamps.24 how_to_obtain: "Create a Deepgram API key. Set `DEEPGRAM_API_KEY` through browser admin at the active `/admin/secrets` route; if browser admin is unavailable, use `/secret set DEEPGRAM_API_KEY \"<api-key>\"` in browser `/chat` or TUI; local console fallback: `hybridclaw secret set DEEPGRAM_API_KEY \"<api-key>\"`."25 - id: assemblyai-api-key26 kind: api_key27 required: false28 secret_ref:29 source: store30 id: ASSEMBLYAI_API_KEY31 scope: AssemblyAI async speech-to-text transcription, diarization, and word timestamps.32 how_to_obtain: "Create an AssemblyAI API key. Set `ASSEMBLYAI_API_KEY` through browser admin at the active `/admin/secrets` route; if browser admin is unavailable, use `/secret set ASSEMBLYAI_API_KEY \"<api-key>\"` in browser `/chat` or TUI; local console fallback: `hybridclaw secret set ASSEMBLYAI_API_KEY \"<api-key>\"`."33metadata:34 source: https://skillmd.com/skills/hybridaione/speech-transcribe35 hybridclaw:36 category: media37 short_description: Provider-agnostic speech-to-text transcripts.38 config:39 speechToText:40 defaultProvider: auto41 tags:42 - speech43 - transcription44 - diarization45 - timestamps46 - audio47 related_roadmap:48 - R21.6949 issue: 99950 stakes_tiers:51 green:52 - provider-list53 - private-transcript54 amber:55 - provider-call56 - transcript-artifact57 red:58 - public-share59 escalation:60 writes: confirm-each61 route: f862 cost_measurement:63 system: UsageTotals64 sub_limit_contract: R21.10065 sub_limit_key: speech-to-text66---6768# Speech Transcribe6970Use the `audio_transcribe` tool when the user asks to transcribe,71caption, diarize, timestamp, or identify speakers in an audio or video clip.7273The tool comes from the `media-tools` plugin. If it is not in your tool list,74tell the user to enable it with `hybridclaw plugin install media-tools`.7576## Workflow77781. Call `audio_transcribe` with `action: "list"` if you need provider79 readiness or the user asks what is configured.802. Pass `audio` as a current attachment filename/ref, `/workspace` path,81 `/discord-media-cache` path, `/uploaded-media-cache` path, or HTTPS media82 URL.833. Use `provider: "auto"` unless the user asks for `openai`, `deepgram`, or84 `assemblyai`, or unless diarization is required. Auto mode honors the tenant85 `skills.speechToText.defaultProvider` config when it is set. Prefer Deepgram86 or AssemblyAI for speaker labels.874. Pass `language` only when the user gives a known language. Omit it for88 provider language detection.895. Set `diarization: true` when the user asks for speaker labels.906. Set `timestamps` to `word`, `segment`, or `none` based on the request.917. Return the structured result fields that matter: transcript text, provider,92 detected language, duration, cost, warnings, and artifact paths.9394The tool owns provider credentials, provider fallback, output schema,95long-audio chunking for local and remote OpenAI uploads when `ffmpeg`/`ffprobe`96are available, transcript artifact persistence, and usage-cost accounting.9798## Output Contract99100The tool returns JSON shaped like:101102```json103{104 "text": "Transcript text",105 "segments": [{ "start": 0, "end": 1.2, "speaker": "speaker_0", "text": "..." }],106 "language": "en",107 "provider": "deepgram",108 "duration_sec": 12.3,109 "cost_usd": 0.001110}111```112113Transcript text and segment JSON are also persisted as private workspace114artifacts. Treat transcripts as operator-private until the user explicitly asks115to share, post, email, or publish them.