at a glance
quickstart
one call, one.wav file:
opus, pcm, mulaw, alaw, or mp3, add audio_format to the
request. see audio formats for exact values and examples.
put the voice in description and your spoken text in text. the full request
schema is in the api reference.
preset voices
description is what builds the voice, so always send it. if it is missing or
empty the api does not reject the request, it falls back to a generic neutral
voice. leave speaker out and mulberry generates a voice to match the description you wrote. add
speaker to pin one of twelve named voices instead:
send
description alongside it, exactly as you would without a speaker:
description, not an error, so a typo sounds like the wrong voice
rather than failing loudly.
deprecating soon. the old numbered preset values still work for now and map
to
ira, siya, aisha and zoya, in that order. they’re going away, so move
to the names above.parameters
getting truncated audio? if the returned speech stops before the end of your
text, the generation hit the token cap. raise
max_new_tokens above the default
2048, up to 8000. the engine clamps anything higher to 8000 rather than
rejecting it.