Analyze prosody

Analyze emotional expression from speech audio.

This endpoint requires authentication. Include your API key in the Authorization header. This is an async job-based endpoint. Submit audio, then poll for results.

How It Works

Submit audio via POST /v1/prosody — returns a job_id
Poll status via GET /v1/prosody/{job_id} — returns queued, in_progress, completed, or failed
Retrieve results via GET /v1/prosody/{job_id}/result — returns 5 emotion metrics

Request Body (multipart/form-data)

Parameter	Type	Required	Description
`file`	binary	Yes	Audio file (WAV, MP3, OGG, FLAC, WebM)
`model`	`valsea-prosody`	Yes
`language`	string	No	BCP-47 language hint (e.g. "en", "zh")
`granularity`	`word` \| `sentence` \| `utterance`	No	Default: `utterance`
`response_format`	`json` \| `verbose_json`	No	Default: `json`

Code Examples

curl -X POST https://api.valsea.ai/v1/prosody \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "model=valsea-prosody" \
  -F "file=@audio.wav"

Response

Submit Response (202)

Field	Type	Description
`job_id`	string	Unique job identifier
`status`	string	`queued`

Minimal (`json`)

Field	Type	Description
`job_id`	string
`status`	string	`completed`
`emotions.frustration`	number	Frustration score (0-1)
`emotions.stress`	number	Stress score (0-1)
`emotions.politeness`	number	Politeness score (0-1)
`emotions.hesitation`	number	Hesitation score (0-1)
`emotions.urgency`	number	Urgency score (0-1)

Verbose (`verbose_json`)

Includes all fields from the minimal response, plus:

Field	Type	Description
`raw_predictions`	array	Raw expression scores per utterance
`raw_predictions[].text`	string	Utterance text
`raw_predictions[].time.begin`	number	Start time (seconds)
`raw_predictions[].time.end`	number	End time (seconds)
`raw_predictions[].emotions`	array	All 48 expression scores

Errors

Status	Description
`401`	Missing or invalid API key
`402`	Insufficient credits
`404`	Job not found or expired
`409`	Job not yet completed
`503`	Service not configured