zooqueen
ae80da5998
Voice is POST /v1/audio/speech and word timings come from POST /v1/audio/transcriptions over the rendered audio. Music, sound effects, images, and video generate through /v1/audio/music, /v1/audio/foley, /v1/images/generations, and /v1/videos/generations. The shared catalog is a listing under media/<type>/ in our own object store, browsed through /v1/s3 — an empty prefix reports itself as empty rather than reading as a bad query. One client (scripts/lib/api.mjs) holds every route and the one credential: a Hanzo IAM token in $HANZO_API_KEY, read from Hanzo KMS. No credential file, no per-repo .env, no per-provider CLI to install or keep current. $HANZO_BASE_URL points the same code at a local Hanzo Engine. Word timings are passed through, never interpolated: a caption cut on invented boundaries reads as correct and drifts against the voice. When the service returns text without them, the word array is empty and the caller is told. Deletes the provider cascade this replaces, its model ladders, and the account linking in telemetry — the install id stands alone.
1.4 KiB
Transcription
Word-level timestamps from audio, through POST /v1/audio/transcriptions.
node <SKILL_DIR>/scripts/transcribe.mjs --input audio.mp3
node <SKILL_DIR>/scripts/transcribe.mjs --input video.mp4 --lang es
node <SKILL_DIR>/scripts/transcribe.mjs --input audio.mp3 --out words.json --json
Language
--lang defaults to en and is sent as the request language. Pass the real
language of the audio: a transcription asked for the wrong language returns
translated text, which silently destroys the original. For mixed-language audio,
name the one you want kept.
Model
--model defaults to whisper. whisper-small is the faster, smaller
alternative. Both are served by our speech service; there is nothing to download
and no first-run build.
Output shape
A flat array of word objects. The id (w0, w1, …) is added during
normalization so caption overrides have a stable reference.
{
"text": "Hello world.",
"words": [
{ "text": "Hello", "start": 0.0, "end": 0.5 },
{ "text": "world.", "start": 0.6, "end": 1.2 }
]
}
words is empty when the service returns text without per-word timings, and
the command says so on stderr. Nothing is interpolated to fill the gap. Callers
that need per-word cuts should treat an empty words as "captions fall back to
line timing", not as a failed transcription.
For caption-quality checks and retry rules, see
captions/transcript-handling.md.