December 11th.Wall-facing intelligenceOfficial announcement,VoxCPM version 1.5 officially onlineThe continuous optimization of the developer ' s development experience has been accompanied by a number of core competency upgrades。
VoxCPM is a 0.5B parameter sizeVoice GenerationThe base model was first released in September this year。
1AI with VoxCPM 1.5 Update highlights:
- High-extract sample audio cloning:The AudioVAE sampling rate has been raised from 16 kHz to 44.1 kHz, and the model is based on high-quality audio, which allows for better cloning and more detailed voices
- Generation efficiency doubled:Subject to an increase in model parameters, VoxCPM 1.5 The only 6.25 token required to generate a second audio, double the previous version, increasing the quality of audio generation while maintaining speed
- Developer friendly:New LoRA and full fine-tuning scripts to support in-depth customization
- Enhancing stability:Reduces audio perjury and optimizes the production of long text audio。
Currently VoxCPM 1.5 models are in Github, Hugging Face Open Source.
- Huggingface: https://huggingface.co/openbmb/VoxCPM1.5
- Github: https://github.com/OpenBMB/VoxCPM
statement:The content of the source of public various media platforms, if the inclusion of the content violates your rights and interests, please contact the mailbox, this site will be the first time to deal with.