elevenlabs-live-vc 🎙️

Live speech to speech bot using Eleven Labs API.

██╗     ██╗██╗   ██╗███████╗    ██╗   ██╗ ██████╗
██║     ██║██║   ██║██╔════╝    ██║   ██║██╔════╝
██║     ██║██║   ██║█████╗█████╗██║   ██║██║
██║     ██║╚██╗ ██╔╝██╔══╝╚════╝╚██╗ ██╔╝██║
███████╗██║ ╚████╔╝ ███████╗     ╚████╔╝ ╚██████╗
╚══════╝╚═╝  ╚═══╝  ╚══════╝      ╚═══╝   ╚═════╝

Description: A live voice-changer utilizing elevenlabs voice-cloning API.
Author: https://github.com/cavoq

Features ✨

• Real-time Voice Transformation - Transform your voice using ElevenLabs' AI voice cloning
• VB-Cable Integration - Automatically routes transformed audio to VB-Cable for use in calls
• Two Recording Modes:
  • Manual Mode (MODE=0) - Press SPACE to start/stop recording
  • Automatic Mode (MODE=1) - Voice Activity Detection (VAD) auto-detects speech
• Configurable Audio Settings - Sample rate, channels, silence threshold, noise reduction

Use Cases 💡

• Discord voice calls with voice changing
• Zoom/Teams meetings
• WhatsApp calls (via WhatsApp Desktop)
• Streaming with OBS
• Any application that supports microphone input

Prerequisites 🧰

1. ElevenLabs Account - Get your API key from ElevenLabs
2. VB-Cable - Download and install from https://vb-audio.com/Cable/
3. FFmpeg - Required for audio processing. Download from https://ffmpeg.org/download.html
4. Python 3.12.x or 3.13.x (3.13.12 recommended) - Download from https://python.org
5. uv (optional, recommended) - Install from https://docs.astral.sh/uv/

Installation 📦

1. Clone the repository

git clone https://github.com/cavoq/elevenlabs-live-vc.git
cd elevenlabs-live-vc

2. Install dependencies

Option A: uv (recommended)

## uv can install Python for you if you don't have 3.13.12 yet
uv python install 3.13.12
uv venv --python 3.13.12
uv sync

Option B: pip

pip install -r requirements.txt

3. Environment Variables

Create a file named .env in the root directory:

# Required - Get from https://elevenlabs.io/app/settings/api-keys
API_KEY=your_elevenlabs_api_key_here

# Required - Get from ElevenLabs Voice Lab
# Go to https://elevenlabs.io/app/voice-lab → Select a voice → Copy Voice ID
VOICE_ID=your_voice_id_here

# Optional - Audio settings
SAMPLE_RATE=48000
CHANNELS=1
SILENCE_THRESHOLD=0.01
VAD_THRESHOLD=0.01
REMOVE_BACKGROUND_NOISE=1
OUTPUT_SAMPLE_RATE=48000
API_SAMPLE_RATE=22050
# Optional - audio output selection
# OUTPUT_DEVICE=5
# OUTPUT_DEVICE_NAME=VB-Cable
# Optional - audio input selection
# INPUT_DEVICE=2
# INPUT_DEVICE_NAME=Microphone

# Optional - VAD (Automatic mode)
VAD_SILENCE_DURATION=0.8
VAD_MIN_RECORDING_DURATION=0.3
VAD_PRE_BUFFER_DURATION=0.5

# Optional - Recording mode
# 0 = Manual (press SPACE to record)
# 1 = Automatic (Voice Activity Detection)
MODE=0

Usage ▶️

Start the Application

Option A: uv

uv run python live-vc.py

Option B: Python

python live-vc.py

Manual Mode (Default)

1. Press SPACE to start recording
2. Speak into your microphone
3. Press SPACE again to stop
4. Wait for processing - transformed voice plays to VB-Cable

Automatic Mode (VAD)

Enable by setting MODE=1 in .env or typing set_mode 1 in the app.

1. Just start speaking - recording begins automatically
2. Stop speaking - after a short silence, recording stops
3. Processing happens automatically
4. Cycle repeats - starts listening again after processing

Commands

Command	Description
set_mode 0	Switch to manual mode (press SPACE)
set_mode 1	Switch to automatic mode (VAD)
get_mode	Show current mode
clear	Clear the screen
quit	Exit the application

Using with Call Applications

Setup for Discord/Zoom/WhatsApp/etc.

1. Install VB-Cable if not already installed
2. Run the voice changer: python live-vc.py
3. In your call app, go to Settings → Audio/Voice
4. Set Microphone/Input to "CABLE Output"
5. Start talking - your transformed voice will be heard by others

Audio Flow

flowchart LR
  A["Your Mic"]
  B["elevenlabs-live-vc"]
  C["VB-Cable Input"]
  D["VB-Cable Output"]
  E["Call App"]
  A --> B --> C --> D --> E

Loading

Docker 🐳

docker build -t el-live-vc .
docker run --env-file .env -it --privileged -v /dev/input:/dev/input el-live-vc

Troubleshooting 🛠️

"No audio recorded" message

• Make sure you're speaking long enough (at least 0.5 seconds)
• Check that your microphone is set as the default Windows input device
• Verify microphone permissions in Windows Settings

VB-Cable not detected

• Ensure VB-Cable is installed correctly
• The app looks for a device containing "CABLE Input" in the name
• Restart the app after installing VB-Cable

Voice not heard in call apps

• Make sure you selected "CABLE Output" as the microphone in your call app
• Check that VB-Cable is not muted in Windows Sound settings
• Verify the app shows "Done! Audio sent to VB-Cable."

API Errors

• Verify your API key is correct in .env
• Check your ElevenLabs account has available credits
• Ensure the Voice ID exists and you have access to it

Configuration Options ⚙️

Environment Variable	Default	Description
API_KEY	(required)	Your ElevenLabs API key
VOICE_ID	(required)	The voice to transform into
SAMPLE_RATE	48000	Audio sample rate in Hz
CHANNELS	1	Number of audio channels (1=mono)
SILENCE_THRESHOLD	0.01	Silence trim threshold (RMS)
VAD_THRESHOLD	0.01	Voice detection threshold (RMS)
REMOVE_BACKGROUND_NOISE	1	1=Enable, 0=Disable
OUTPUT_SAMPLE_RATE	48000	Playback device sample rate in Hz
API_SAMPLE_RATE	22050	API output sample rate in Hz (PCM)
OUTPUT_DEVICE	(optional)	Audio output device index
OUTPUT_DEVICE_NAME	(optional)	Output device name substring
INPUT_DEVICE	(optional)	Audio input device index
INPUT_DEVICE_NAME	(optional)	Input device name substring
VAD_SILENCE_DURATION	0.8	Seconds of silence before auto-stop
VAD_MIN_RECORDING_DURATION	0.3	Minimum recording length in seconds
VAD_PRE_BUFFER_DURATION	0.5	Pre-buffer audio in seconds
MODE	0	0=Manual, 1=Automatic (VAD)

License 📄

GNU General Public License v3.0 - See LICENSE for details.

Credits 🙌

• Author: cavoq
• Additional Contributors: ayeantics
• ElevenLabs: https://elevenlabs.io
• VB-Audio: https://vb-audio.com