Choose or clone a voice

Every speech request uses one of two voice paths.

Use a voice ID

Pass default or an ID returned by client.voices.list():

speech = client.audio.speech.create(
    model="teen-v1",
    voice="default",
    input="Welcome to the show.",
    response_format="mp3",
)

Clone from reference audio

from pathlib import Path

speech = client.audio.speech.create(
    model="teen-v1",
    input="Speak using this reference voice.",
    reference_audio=Path("reference.wav"),
    reference_audio_format="wav",
    response_format="mp3",
)

Reference audio may be WAV, MP3, FLAC, or OGG. The SDK accepts bytes, paths, or valid base64 text.

Do not combine a saved voice ID with reference audio. Voice descriptions and reference transcripts are not speech parameters; unsupported fields fail instead of changing the generated audio.