Legacy Endpoint Migration

If you have already integrated the legacy /api/v1 task model, you need to migrate to the current /v1 audio endpoints. The two are not the same contract.

Path migration

Capability	Legacy path	New path
Single-speaker speech generation	POST /api/v1/tts	POST /v1/audio/speech
Multi-speaker speech generation	POST /api/v1/ttsd	POST /v1/audio/speech/speakers
Audio transcription	POST /api/v1/transcriptions	POST /v1/audio/transcriptions
Task query	GET /api/v1/tasks/{client_task_id}	GET /v1/audio/tasks/{task_id}
File upload	POST /api/v1/files/upload	POST /v1/files
Create a voice	POST /api/v1/voice/clone	POST /v1/audio/voices

Field migration

Legacy field	New field / new behavior
`model_id` / `model_version`	`model`
`client_task_id`	`task_id` returned by the server
`callback_url`	`webhook_url`
`voice_url` / `voice_data`	Create a `voice_id` first, then use `voice` or `voice_id` in TTS
`audio_data`	Not supported; ASR uses `file` / `file_id` / `url` / `audio_url`
`response_format` disabled	`response_format` is now supported, defaulting to `mp3`