Voice Cloning

Instant Voice Clone (Web UI)

View as Markdown

Clone any voice from 5-15 seconds of reference audio using the web console. No code required. Upload a sample, preview the clone, and get a voice_id to use in your TTS API calls immediately.

Prerequisites

For recording guidelines, see Voice Cloning Best Practices.

Step 1: Navigate to Voice Cloning

Log in to app.smallest.ai and select Voice Cloning from the left sidebar under Playground.

Smallest AI dashboard with Voice Cloning highlighted in the left sidebar

Step 2: Start a New Clone

Click the Create button. If this is your first clone, the button is displayed in the center of the page. Otherwise, use the Create button in the top-right corner.

Voice Cloning page showing the Create button

A guide dialog appears with best practices for reference audio:

  • Use a quiet environment to avoid background noise, hiss, or rumble.
  • Record clean, high-quality reference audio of your natural speaking voice.
  • Speak normally. The model captures timbre, accent, emotional tone, rhythm, and pacing automatically.
Instant Voice Cloning guide showing recording best practices

Click Next to proceed.

Step 3: Provide Reference Audio

Provide an audio clip of 5 to 15 seconds. Two options are available:

  1. Record Audio — Record a live clip directly in the browser. Allow microphone access when prompted.
  2. Upload or drag and drop — Upload a pre-recorded file. Supported formats: .mp3, .wav, .mp4.

The console provides a suggested text prompt to read aloud if you choose to record live.

Add audio for cloning dialog showing Record Audio and upload options

Step 4: Preview and Select Language

After uploading or recording, preview your audio using the playback controls. Select the target language(s) for the clone. English is selected by default; Hindi is also available.

For best results, record reference audio in the same language as your intended output. See Voice Cloning Best Practices for multi-lingual cloning guidance.

Language selection dialog showing English selected and Hindi available with audio preview

Click Next to proceed.

Step 5: Listen to Clone Previews

The platform generates sample audio clips using your cloned voice. Listen to the previews to verify the voice matches your expectations. If the quality is not satisfactory, click Back to re-record or upload different reference audio.

Voice clone preview dialog showing three generated audio samples with playback controls

Click Next to proceed.

Step 6: Name and Save

Provide a Name, Tags, and optional Description for the cloned voice. Accept the terms and conditions, then click Finish to save.

Create a Voice Clone dialog showing name, tags, and description fields with Finish button

Using Your Cloned Voice

After saving, the cloned voice appears in your Voice Cloning dashboard with its voice_id, type, tags, accent, and creation date.

Voice Cloning dashboard listing cloned voices with name, type, tags, accent, and creation date

Click the Use button next to your clone to open the Text to Speech playground with your cloned voice pre-selected.

Voice Cloning dashboard with Use button highlighted for the cloned voice

The TTS playground opens with your cloned voice selected under Voice, the model set to Lightning v3.1, and controls for pronunciation language, speed, and sample rate. Enter text and click Generate to synthesize speech with your cloned voice.

Text to Speech playground showing cloned voice selected with model, language, speed, and sample rate controls

Using the Clone via API

Pass the voice_id from your dashboard in any TTS API call:

$curl -X POST "https://api.smallest.ai/waves/v1/lightning-v3.1/get_speech" \
>  -H "Authorization: Bearer $SMALLEST_API_KEY" \
>  -H "Content-Type: application/json" \
>  -d '{
>    "text": "Hello, this is my cloned voice.",
>    "voice_id": "voice_xsJ0rrvuqR"
>  }' \
>  -o output.pcm

For programmatic voice cloning (create clones via code), see Voice Cloning via API.

Need help? Contact support@smallest.ai or ask on Discord.