Voice cloning

Voice cloning turns a short reference clip of a voice plus some text into speech spoken in that voice. It's a stateless call — you upload the reference audio with each request, so there's nothing to pre-register.

• Endpoint: POST /v1/audio/voice-clone
• Model: viitor-voice-nar
• Request: multipart/form-data
• Response: an audio stream (the cloned speech)

Request fields

Field	Type	Required	Notes
model	text	yes	viitor-voice-nar
input	text	yes	The text to speak in the cloned voice
reference_audio	file	yes	A short clip (a few seconds) of the voice to clone — wav recommended
language	text	no	Language hint for the output

Keep the reference clip short and clean — a few seconds of clear speech works best. Uploads are capped at 32 MB.

Example

curl https://api.ecohash.com/v1/audio/voice-clone \
  -H "Authorization: Bearer $ECOLINK_API_KEY" \
  -F model=viitor-voice-nar \
  -F input="Hello — this sentence is spoken in the cloned voice." \
  -F language=en \
  -F reference_audio=@sample.wav \
  --output cloned.wav

The response body is the generated audio. Each response includes an Inference-Id header — a per-request identifier useful for support and debugging.

Billing

Voice cloning is billed as speech synthesis, at the model's speech rate. See Balance & transactions for where usage appears.

Related

• Speech editing — edit words in existing audio while keeping the same voice (same model)
• Text-to-speech — synthesize speech with the platform's standard voices
• Voice-clone & editing API reference