| import torch, soundfile as sf, librosa, numpy as np |
| from vibevoice.processor.vibevoice_processor import VibeVoiceProcessor |
| from vibevoice.modular.modeling_vibevoice_inference import VibeVoiceForConditionalGenerationInference |
|
|
| |
| voice, sr = sf.read("path/to/voice_sample.wav") |
| if voice.ndim > 1: voice = voice.mean(axis=1) |
| if sr != 24000: voice = librosa.resample(voice, sr, 24000) |
|
|
| processor = VibeVoiceProcessor.from_pretrained("tarun7r/vibevoice-hindi-7b") |
| model = VibeVoiceForConditionalGenerationInference.from_pretrained( |
| "tarun7r/vibevoice-hindi-7b", torch_dtype=torch.bfloat16 |
| ).to("cuda").eval() |
| model.set_ddpm_inference_steps(5) |
|
|
| inputs = processor(text=["Speaker 0: Hello!\nSpeaker 1: Hi there!"], |
| voice_samples=[[voice]], return_tensors="pt") |
| audio = model.generate(**inputs, cfg_scale=1.3, |
| tokenizer=processor.tokenizer).speech_outputs[0] |
| sf.write("output.wav", audio.cpu().numpy().squeeze(), 24000) |
