Extensions/ComfyUI-FL-VoxCPM
ComfyUI Extension

ComfyUI-FL-VoxCPM

A ComfyUI extension with 8 custom nodes.

By filliptm·Created 7 months ago·Updated 4 months ago· 32
filliptm/ComfyUI-FL-VoxCPM
Nodes8
On cloudLocal install
CategoryFL/VoxCPM, FL/VoxCPM/Training
Stars32
Updated4 months ago

Nodes (8)

FL VoxCPM Audio Crop

Trim reference audio before you clone a voice — it matters more than you think

FL/VoxCPM
FL VoxCPM Dataset Maker

The two-minute dataset builder for VoxCPM voice LoRAs

FL/VoxCPM/Training
FL VoxCPM LoRA Trainer

Train a voice LoRA inside ComfyUI, with a live loss chart and audio samples

FL/VoxCPM/Training
FL VoxCPM Train Config

V1 LoRA training settings — what actually moves the needle

FL/VoxCPM/Training
FL VoxCPM Transcribe

Free transcriptions, because your cloned voice needs to know what was said

FL/VoxCPM
FL VoxCPM TTS

Fast, accurate local TTS with voice cloning — the one people forget about

FL/VoxCPM
FL VoxCPM V2 Train Config

The V2 training config that ships with sane OpenBMB defaults

FL/VoxCPM/Training
FL VoxCPM V2 TTS

Design a voice from a sentence, clone one from a clip — one node, five modes

FL/VoxCPM
Readme

FL VoxCPM

Text-to-speech nodes for ComfyUI powered by OpenBMB's VoxCPM model family. Features V1 and V2 models, voice cloning, voice design from text descriptions, LoRA fine-tuning with real-time training dashboard, and audio utility nodes.

VoxCPM Patreon

Workflow Preview

Features

  • VoxCPM V2 - 2B parameter model, 48kHz studio-quality audio, 30 languages
  • VoxCPM V1.5 - 44.1kHz high-fidelity TTS with 800M parameters
  • Voice Design - Create custom voices from natural language descriptions (V2)
  • Voice Cloning - Clone any voice from a short audio reference
  • Controllable Cloning - Clone a voice and modify its style/emotion (V2)
  • Ultimate Cloning - Maximum fidelity cloning with reference + continuation audio (V2)
  • LoRA Training - Fine-tune custom voices with real-time dashboard (loss chart, progress, validation audio)
  • Auto Transcription - Built-in Whisper integration for generating reference text
  • Audio Crop - Trim audio to specific time ranges

Nodes

| Node | Description | |------|-------------| | FL VoxCPM TTS | V1/V1.5 text-to-speech with optional voice cloning and LoRA | | FL VoxCPM V2 TTS | V2 TTS with Voice Design, Voice Cloning, Controllable Cloning, and Ultimate Cloning modes | | FL VoxCPM Train Config | V1 training hyperparameters (learning rate, LoRA rank, etc.) | | FL VoxCPM V2 Train Config | V2 training hyperparameters (48kHz defaults, rank 32, alpha 32) | | FL VoxCPM Dataset Maker | Create training dataset from audio + transcript files | | FL VoxCPM LoRA Trainer | Unified trainer for V1 and V2 with real-time dashboard and validation audio | | FL VoxCPM Transcribe | Transcribe audio to text using Whisper (useful for prompt_text / reference_text) | | FL VoxCPM Audio Crop | Trim audio to specific start/end times |

Installation

ComfyUI Manager

Search for "FL VoxCPM" and install.

Manual

cd ComfyUI/custom_nodes
git clone https://github.com/filliptm/ComfyUI-FL-VoxCPM.git
cd ComfyUI-FL-VoxCPM
pip install -r requirements.txt

Quick Start

V2 Voice Design (No Reference Audio Needed)

  1. Add FL VoxCPM V2 TTS node
  2. Set mode to Voice Design
  3. Enter voice description in control (e.g. "young woman, warm and gentle voice")
  4. Enter target text, queue prompt

V2 Voice Cloning

  1. Add FL VoxCPM V2 TTS + Load Audio nodes
  2. Set mode to Voice Cloning
  3. Connect reference audio to reference_audio
  4. Enter target text, queue prompt

V1 Text-to-Speech

  1. Add FL VoxCPM TTS node
  2. Enter text, queue prompt
  3. Model downloads automatically on first run

V1 Voice Cloning

  1. Add FL VoxCPM TTS + Load Audio nodes
  2. Connect reference audio to prompt_audio
  3. Use FL VoxCPM Transcribe to generate transcript, or enter it manually in prompt_text
  4. Enter target text, queue prompt

LoRA Training

  1. Prepare folder with .wav/.mp3/.flac files + matching .txt transcripts
  2. FL VoxCPM Dataset Maker -> FL VoxCPM Train Config (or V2) -> FL VoxCPM LoRA Trainer
  3. Set validation_text to hear samples at each checkpoint
  4. Monitor training in the real-time dashboard (loss chart, progress bar, audio samples)
  5. LoRA saves to ComfyUI/models/loras/VoxCPM/

Models

| Model | Parameters | Sample Rate | Languages | Notes | |-------|-----------|-------------|-----------|-------| | VoxCPM2 | 2B | 48kHz | 30 | Recommended - Voice Design, Controllable Cloning | | VoxCPM1.5 | 800M | 44.1kHz | 2 | Stable, high quality | | VoxCPM-0.5B | 500M | 16kHz | 2 | Legacy, lightweight |

Models auto-download from HuggingFace to ComfyUI/models/tts/VoxCPM/ on first use.

V2 Modes

| Mode | Required Inputs | Description | |------|----------------|-------------| | Voice Design | text + control | Generate speech from a voice description - no audio needed | | Voice Cloning | text + reference_audio | Clone a voice from reference audio | | Controllable Cloning | text + reference_audio + control | Clone a voice and modify its style/emotion | | Ultimate Cloning | text + reference_audio + reference_text + prompt_audio + prompt_text | Maximum fidelity with full context |

Key Parameters

  • cfg_value (2.0) - Guidance scale. Higher = closer to reference voice, lower = more natural
  • inference_timesteps (10) - Diffusion steps. More = better quality, slower
  • lora_name - Select trained LoRA from dropdown (auto-detects rank)
  • control (V2 only) - Natural language voice description for design/style control

Dataset Format (Training)

audio_folder/
  sample1.wav
  sample1.txt    # contains transcript of sample1.wav
  sample2.mp3
  sample2.txt
  ...

Supported audio formats: .wav, .mp3, .flac

Requirements

  • Python 3.9+
  • 16GB RAM minimum (32GB+ recommended for training)
  • NVIDIA GPU with 12GB+ VRAM recommended (CPU and Mac MPS supported for V1 inference)
  • V2 model requires ~10GB VRAM for inference, ~20GB for training

License

Apache-2.0 - Based on OpenBMB/VoxCPM