Chatterbox-Nano-TTS-FlashVocoder-fp16
akashmjn/Chatterbox-Nano-TTS-fp16 with the flow weights of its S3Gen vocoder replaced by those from ResembleAI/chatterbox-flash.
Flash's s3gen.safetensors is the same meanflow S3Gen architecture as
Turbo/Nano's, differing only in the 1122 flow.* tensors; mel2wav,
speaker_encoder and the tokenizer are byte-identical. Only those s3gen.*
flow tensors are changed here. The Nano T3 backbone, voice encoder, tokenizer
files and bundled conds.safetensors are unchanged from the base checkpoint.
Loads like the base checkpoint, via the chatterbox_turbo model in
akashmjn/mlx-audio (variant: nano).
- Downloads last month
- 45
Model size
0.4B params
Tensor type
F16
·
Hardware compatibility
Log In to add your hardware
Quantized
Model tree for akashmjn/Chatterbox-Nano-TTS-FlashVocoder-fp16
Base model
ResembleAI/chatterbox-flash