ai-agents-device-tensor-handli-when-using-voice-preset-with-barkprocessor-the-his-ac4ee974

THEN Manually load the voice_preset from the .npz file into a dictionary, convert the 'history_prompt' tensor to the target device (e.g., CUDA) before moving the inputs object to the same device. This ensures all tensors are on the same device for generation.

with numpy.load('bark/assets/prompts/en_speaker_6.npz') as data:
    voice_preset = {
        'fine_prompt': data['fine_prompt'],
        'coarse_prompt': data['coarse_prompt'],
        'semantic_prompt': data['semantic_prompt']
    }
inputs = processor(text_prompt, voice_preset=voice_preset)
inputs['history_prompt'] = inputs['history_prompt'].to(device)
inputs.to(device)