Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.auvy.ai/llms.txt

Use this file to discover all available pages before exploring further.

​
Voice API

The voice plane configures speech-to-text custom vocabulary for batch transcription on agent-end. It does not accept raw audio bytes — ingest via chat attachments or assets, then use Transcription.

​
Prewarm vocabulary

Loads and caches merged vocabulary (core + vertical + workspace/brain terms) for the current auth context. 
POST /v1/voice/stt-vocabulary/prewarm

const auvy = await fromApiKey()
const { term_count } = await auvy.voice.prewarmSttVocabulary()
Body validates workspace_id and optional brain_id against auth. Call after brain switch or before heavy transcription batches.

​
Defaults (static layers)

GET /v1/voice/stt-vocabulary/defaults

const defaults = await auvy.voice.sttVocabularyDefaults()
// { vertical_id, core, vertical, merged_static }
Returns platform static layers for the configured STT vertical — useful for debugging term coverage without prewarm side effects.

​
Batch STT workflow

  1. Upload audio via POST /v1/chat/attachments or POST /v1/assets
  2. Optional: POST /v1/voice/stt-vocabulary/prewarm
  3. POST /v1/transcribe with { source: { type, id } }
  4. Poll job or auvy.transcription.waitForResult(job_id)
See Transcription API.

​
SDK namespace

auvy.voice — two methods: prewarmSttVocabulary, sttVocabularyDefaults.