admin/VoxCPM
1
0
Fork 0
mirror of https://github.com/OpenBMB/VoxCPM synced 2025-12-12 03:48:12 +00:00
Code Issues Packages Projects Releases Wiki Activity

Update: default no denoise & normalize

Browse Source
This commit is contained in:
Labmem-Zhouyx
2025-12-05 22:16:27 +08:00
parent 6a5e713698
commit b1f7593ae0
2 changed files with 9 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
README.md
View File

@@ -76,7 +76,7 @@ By default, when you first run the script, the model will be downloaded automati
from huggingface_hub import snapshot_download from huggingface_hub import snapshot_download
snapshot_download("openbmb/VoxCPM-0.5B") snapshot_download("openbmb/VoxCPM-0.5B")
``` ```
- Download ZipEnhancer and SenseVoice-Small. We use ZipEnhancer to enhance speech prompts and SenseVoice-Small for speech prompt ASR in the web demo. - Download ZipEnhancer and SenseVoice-Small. We use ZipEnhancer to enhance speech prompts and SenseVoice-Small for speech prompt ASR in the web demo.
``` ```
from modelscope import snapshot_download from modelscope import snapshot_download
snapshot_download('iic/speech_zipenhancer_ans_multiloss_16k_base') snapshot_download('iic/speech_zipenhancer_ans_multiloss_16k_base')
@@ -98,8 +98,8 @@ wav = model.generate(
prompt_text=None, # optional: reference text prompt_text=None, # optional: reference text
cfg_value=2.0, # LM guidance on LocDiT, higher for better adherence to the prompt, but maybe worse cfg_value=2.0, # LM guidance on LocDiT, higher for better adherence to the prompt, but maybe worse
inference_timesteps=10, # LocDiT inference timesteps, higher for better result, lower for fast speed inference_timesteps=10, # LocDiT inference timesteps, higher for better result, lower for fast speed
normalize=True, # enable external TN tool, but will disable native raw text support normalize=False, # enable external TN tool, but will disable native raw text support
denoise=True, # enable external Denoise tool, but it may cause some distortion and restrict the sampling rate to 16kHz denoise=False, # enable external Denoise tool, but it may cause some distortion and restrict the sampling rate to 16kHz
retry_badcase=True, # enable retrying mode for some bad cases (unstoppable) retry_badcase=True, # enable retrying mode for some bad cases (unstoppable)
retry_badcase_max_times=3, # maximum retrying times retry_badcase_max_times=3, # maximum retrying times
retry_badcase_ratio_threshold=6.0, # maximum length restriction for bad case detection (simple but effective), it could be adjusted for slow pace speech retry_badcase_ratio_threshold=6.0, # maximum length restriction for bad case detection (simple but effective), it could be adjusted for slow pace speech
@@ -134,14 +134,14 @@ voxcpm --text "VoxCPM is an innovative end-to-end TTS model from ModelBest, desi
--prompt-audio path/to/voice.wav \ --prompt-audio path/to/voice.wav \
--prompt-text "reference transcript" \ --prompt-text "reference transcript" \
--output out.wav \ --output out.wav \
--denoise # --denoise
# (Optinal) Voice cloning (reference audio + transcript file) # (Optinal) Voice cloning (reference audio + transcript file)
voxcpm --text "VoxCPM is an innovative end-to-end TTS model from ModelBest, designed to generate highly expressive speech." \ voxcpm --text "VoxCPM is an innovative end-to-end TTS model from ModelBest, designed to generate highly expressive speech." \
--prompt-audio path/to/voice.wav \ --prompt-audio path/to/voice.wav \
--prompt-file "/path/to/text-file" \ --prompt-file "/path/to/text-file" \
--output out.wav \ --output out.wav \
--denoise # --denoise
# 3) Batch processing (one text per line) # 3) Batch processing (one text per line)
voxcpm --input examples/input.txt --output-dir outs voxcpm --input examples/input.txt --output-dir outs
@@ -149,7 +149,7 @@ voxcpm --input examples/input.txt --output-dir outs
voxcpm --input examples/input.txt --output-dir outs \ voxcpm --input examples/input.txt --output-dir outs \
--prompt-audio path/to/voice.wav \ --prompt-audio path/to/voice.wav \
--prompt-text "reference transcript" \ --prompt-text "reference transcript" \
--denoise # --denoise
# 4) Inference parameters (quality/speed) # 4) Inference parameters (quality/speed)
voxcpm --text "..." --output out.wav \ voxcpm --text "..." --output out.wav \

6
src/voxcpm/core.py
View File

@@ -40,7 +40,7 @@ class VoxCPM:
@classmethod @classmethod
def from_pretrained(cls, def from_pretrained(cls,
hf_model_id: str = "openbmb/VoxCPM-0.5B", hf_model_id: str = "openbmb/VoxCPM1.5",
load_denoiser: bool = True, load_denoiser: bool = True,
zipenhancer_model_id: str = "iic/speech_zipenhancer_ans_multiloss_16k_base", zipenhancer_model_id: str = "iic/speech_zipenhancer_ans_multiloss_16k_base",
cache_dir: str = None, cache_dir: str = None,
@@ -107,8 +107,8 @@ class VoxCPM:
inference_timesteps : int = 10, inference_timesteps : int = 10,
min_len : int = 2, min_len : int = 2,
max_len : int = 4096, max_len : int = 4096,
normalize : bool = True, normalize : bool = False,
denoise : bool = True, denoise : bool = False,
retry_badcase : bool = True, retry_badcase : bool = True,
retry_badcase_max_times : int = 3, retry_badcase_max_times : int = 3,
retry_badcase_ratio_threshold : float = 6.0, retry_badcase_ratio_threshold : float = 6.0,