Audio（OpenAI 標準パス別名）
Estimate speech cost - alias
Alias of POST /v1/tts/estimate.
Text to speech - alias
Alias of POST /v1/tts/generate.
Text to speech - OpenAI path
Alias of POST /v1/tts/speech on the standard OpenAI path.
Transcribe audio - OpenAI path
Alias of POST /v1/tts/transcriptions on the standard OpenAI path. This is the recommended path.
List voices - OpenAI path
Alias of GET /v1/tts/voices. Requires a key. The custody array lists the voices your account has entrusted.
Entrust your own voice
Entrust your own voice. Send a recording together with consent and identity verification. The recording is transcribed and checked against the passphrase and the confirmation number, a voice is created, and the call returns 202 with status pending. The voice cannot be used for speech until AICU has listened to the recording and activated it. A request without consent and verification gets 400; if the statement was not read, 422. Only keys of the account that created the voice can use it.
Revoke an entrusted voice
Revoke an entrusted voice. The voice is also deleted upstream; the record is kept with status revoked.
Get an entrusted voice
Get one entrusted voice. Only voices of your own account; any other id returns 404.
Issue a voice verification statement
Issue the statement to read for identity verification: three passphrase words and a confirmation number. Valid for 24 hours and for one use. Read it at the start of the recording.
Get a verification statement status
Status of a verification statement: whether the speaker has confirmed consent by e-mail, when it expires, and whether it has been used. Owner only.
Voice custody consent text
The canonical consent text for entrusting your own voice: version, full text and its SHA-256.
Reading dictionary - OpenAI path
Alias of GET /v1/tts/yomi. Requires a key.