public synthesize(notnull text as String, notnull voice as String = "", notnull language as String = "auto", notnull instructions as String = "") as AudioClip
Produces a complete independent mono Float32 clip without playback. Memory is O(output samples); use stream() for bounded consumption. Concatenation copies every block once, not the whole preceding utterance.
| Parameter | Description |
|---|---|
text | Nonempty text accepted by the checkpoint. |
voice | Required built-in voice name. |
language | BCP 47 language code, or auto. |
instructions | Optional style instruction, only on compatible checkpoints. |
Complete audio at the checkpoint's output frequency.
clip = speech.synthesize("Bonjour", voice="ryan", language="fr")