Skip to content
Jasiam
Get started

Integration reference · issued-key access

Jasiam Voice API

Narration, selectable voices, reusable assignments, and resumable long-form audio for connected applications.

Access and endpoints

Existing integrations use an issued client key, kept on their server, in an Authorization: Bearer header. There is no anonymous generation or self-service key issuance on this site.

https://jasiam.lovable.app

GET /api/public/voice/voices
Retrieve the supported voice catalog and default.
POST /api/public/voice/preview
Audition a selected voice with optional text up to 300 characters. Requires a key; generation is metered.
POST /api/public/voice/speech
Narrate text with a stable requestId and schoolId; optional voice, courseId, lessonId, and caller segments.
GET / PUT / DELETE /api/public/voice/assignments
Read, set, or remove school, course, and lesson voice defaults.

Wait on Generating

  1. Send the full request with a stable requestId. Keep the entire request unchanged when resuming.
  2. A 202 response with status Generating is progress. Wait retryAfterSec (normally 5 seconds), then repeat the same POST.
  3. Allow at least 90 seconds per call. On a network timeout, retry the same request rather than create a new ID.
  4. A 200 response with status Complete returns audioUrl and durationSec. Finished internal chunks are reused during resume.
  5. Handle 4xx errors explicitly. Back off on rate limits; transient server failures may be retried with the same request.

Optional caller segment timing

Send segments as an ordered array of id and text, optionally with startChar and endChar. When all offsets and full text are supplied, each segment is checked against its source slice; gaps between ranges are allowed. Only segment text is narrated in this mode. Completed responses aggregate real audio durations into ranges with the caller’s IDs. This is segment timing, not word-level forced alignment.

Existing text-only requests remain supported. Keep IDs, text, and offsets stable across retries. Your application owns the selection of narratable content.

Explore Jasiam Voice