Create a Voice Clone | Smallest AI Docs

import requests
url = "https://api.smallest.ai/waves/v1/voice-cloning"
files = { "file": "open('emma_watson_sample.wav', 'rb')" }
payload = {
    "displayName": "Emma Watson Clone",
    "description": ,
    "accent": ,
    "tags": ,
    "language": ,
    "model": 
}
headers = {"Authorization": "Bearer <BearerAuth>"}
response = requests.post(url, data=payload, files=files, headers=headers)
print(response.json())

{
  "message": "Voice clone created successfully",
  "data": {
    "voiceId": "voice_dLP5T67Qw7",
    "displayName": "Emma Watson Clone",
    "model": "lightning-v3.1",
    "status": "pending",
    "language": "en",
    "audioFileNames": [
      "emma_watson_sample_1.mp3"
    ],
    "createdAt": "2024-01-15T09:30:00Z",
    "organizationId": "org_9f8b7c6d5e4a3b2c1d0e",
    "samples": [
      {
        "text": "Hello, this is a sample of the cloned voice.",
        "audioUrl": "https://api.smallest.ai/samples/voice_dLP5T67Qw7/sample1.mp3"
      }
    ]
  }
}

Create an instant voice clone in a single call. Defaults to lightning-v3.1.

Authentication

AuthorizationBearer

Header authentication of the form Bearer <token>

Request

This endpoint expects a multipart form containing a file.

displayNamestringRequired1-500 characters

Human-readable name for the voice clone.

filefileRequired

Audio file to clone from. Supported MIME types: audio/mpeg, audio/mpeg-3, audio/wav, audio/wave, audio/webm, video/webm, audio/mp4, video/mp4. Maximum size: 5 MB.

descriptionstringOptional

Optional longer description for the voice clone.

accentstringOptional

Optional accent tag (e.g. “general”, “indian”).

tagsstringOptional

Optional comma-separated list of tags. Server splits on commas and trims whitespace ("en, tone-test" → ["en", "tone-test"]).

languagestringOptional

Primary language the clone will be used for. Optional, but strongly recommended — set it to the language of your reference audio. When a TTS request later uses language: "auto", the server falls back to this value, so setting it now avoids silent language mismatches at inference time.

Must be one of the languages supported by lightning-v3.1 (e.g. en, hi, multi). The server validates and rejects unsupported codes with a 400.

Primary language the clone will be used for. Optional, but **strongly recommended** — set it to the language of your reference audio. When a TTS request later uses `language: "auto"`, the server falls back to this value, so setting it now avoids silent language mismatches at inference time. Must be one of the languages supported by `lightning-v3.1` (e.g. `en`, `hi`, `multi`). The server validates and rejects unsupported codes with a 400.

modelenumOptionalDefaults to lightning-v3.1

Voice cloning model. Defaults to lightning-v3.1. lightning-v2 is accepted by the schema for historical reasons but is deprecated — the server returns 400 with "Voice cloning for lightning-v2 is deprecated. Please use lightning-v3.1".

Allowed values:

Response

Voice clone created. Includes pre-generated sample clips of the new voice.

messagestring

dataobject

Errors

400

Bad Request Error

401

Unauthorized Error

500

Internal Server Error