Format transcript

Transform a raw transcript into a structured document format such as meeting minutes, sales summaries, action items, subtitles, and more.

This endpoint requires authentication. Include your API key in the Authorization header.

Request Body

Parameter	Type	Required	Description
`model`	`valsea-format`	Yes	Model name. Always `valsea-format`.
`transcript`	string	Yes	The transcript to format.
`semantic_tags`	array	No	Optional context — local expressions and meanings to guide formatting. Default: `[]`.
`semantic_tags[].tag`	string	No	Semantic tag category (e.g. `price_objection`).
`semantic_tags[].phrase`	string	No	The local phrase the tag applies to.
`semantic_tags[].meaning`	string	No	Plain-English meaning of the phrase, used as context.
`output_type`	`meeting_minutes` \| `sales_summary` \| `service_log` \| `subtitles`	Yes	The desired output document format.
`stream`	boolean	No	Enable streaming response.

Code Examples

curl -X POST https://api.valsea.ai/v1/formatting \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "valsea-format",
    "transcript": "Your transcript text here...",
    "output_type": "meeting_minutes"
  }'

Response

The response shape depends on the requested output_type.

`meeting_minutes`

Field	Type	Description
`type`	string	Always `meeting_minutes`.
`summary`	string	Brief 1-2 sentence overview of the meeting.
`agenda_items`	array of strings	Agenda topics covered.
`action_items`	array of `{ description, owner?, deadline? }`	Action items, with optional owner and deadline.
`decisions`	array of strings	Decisions made during the meeting.
`key_discussions`	array of strings	Key discussion points.
`notes`	string	Additional free-form notes.
`generated_at`	string	ISO timestamp.

`sales_summary`

Field	Type	Description
`type`	string	Always `sales_summary`.
`customer_sentiment`	string	`positive` \| `mixed` \| `negative`
`sentiment_reasoning`	string	Why that sentiment was assigned.
`objections`	array of `{ phrase, interpretation, suggested_response }`	Customer objections and how to respond.
`purchase_signals`	array of `{ phrase, interpretation }`	Detected buying signals.
`next_steps`	array of strings	Recommended next steps.
`key_quotes`	array of strings	Notable customer quotes.
`deal_stage`	string	`Discovery` \| `Qualification` \| `Proposal` \| `Negotiation` \| `Closing`
`generated_at`	string	ISO timestamp.

`service_log`

Field	Type	Description
`type`	string	Always `service_log`.
`issue_category`	string	`Technical` \| `Billing` \| `Account` \| `Product` \| `Other`
`issue_summary`	string	Summary of the customer's issue.
`customer_frustration_level`	string	`low` \| `medium` \| `high`
`resolution_status`	string	`Resolved` \| `Pending` \| `Escalated`
`resolution_summary`	string	How the issue was (or will be) resolved.
`root_cause`	string	Identified root cause.
`follow_up_required`	boolean	Whether follow-up is needed.
`follow_up_actions`	array of strings	Recommended follow-up actions.
`semantic_flags`	array of strings	Notable semantic flags from the conversation.
`generated_at`	string	ISO timestamp.

`subtitles`

Field	Type	Description
`type`	string	Always `creator_subtitles`.
`srt`	string	SRT-formatted subtitle text.
`generated_at`	string	ISO timestamp.

Fallback: non-JSON AI output

If the AI's output isn't valid JSON, the endpoint still returns 200, with:

Field	Type	Description
`type`	string	Echoes the requested `output_type`.
`raw_content`	string	The unparsed text returned by the AI.

Errors

Status	Description
`400`	Invalid request — missing `transcript`, invalid `output_type`, or unsupported `model`
`401`	Missing or invalid API key
`402`	Insufficient credits
`429`	Rate limit exceeded
`500`	Format service not configured, or empty AI response
`502`	Upstream AI service error