Local Ai
VoxCPM TTS model + LoRa training abilities right in Comfy
ComfyUI-VoxCPM is a custom node that integrates VoxCPM — a novel tokenizer-free Text-to-Speech system that models speech in a continuous space — directly into ComfyUI's visual workflow environmen...
ComfyUI-VoxCPM is a custom node that integrates VoxCPM — a novel tokenizer-free Text-to-Speech system that models speech in a continuous space — directly into ComfyUI's visual workflow environment, enabling context-aware speech generation and true-to-life zero-shot voice cloning . It supports loading fine-tuned LoRA checkpoints to apply specific voice styles, as well as native LoRA training of custom voice models directly within ComfyUI using dedicated training nodes. Users can fine-tune VoxCPM v1.5 models via LoRA directly inside ComfyUI, allowing them to clone voices or adapt the model's style using a small dataset of audio samples.
Related
- Does anyone have a good example dataset for an Illustrious character Lora that they’re willing to provide?
- kugel-2 model (VibeVoice finetune) repo is gone. Does anyone know why?
- ControlNet vs LoRA
- OmniVoice: multilingual local TTS with 600+ languages, voice cloning, and an OpenAI-compatible server
- ComfyUI - disappearing workflows
- ComfyUI-ConnectTheDots - Connect compatible nodes without scrolling across your graph
Source: local-ai