Vantage-Step-Audio-EditX

EditXModelLoader

ComfyUI node: loads all EditX models once and returns them as a dictionary. Output type: "EDITX_MODELS" – a plain Python dict containing: { "encoder": StepAudioTokenizer, "tts_engine": StepAudioTTS, "model_path": str, }

Vantage/Step Audio EditX · 0 inputs · 3 parameters · 1 output

EditXMultiVoiceCloner

ComfyUI node: Voice cloning using loaded EditX models. Uses StepAudioTTS.clone_from_tensor(prompt_audio_tensor, ..., prompt_text, target_text) Supports: [speaker1]Hello there... [speaker2][happy][serious][faster]Reply... [pause]300 -> insert 300 ms silence

Vantage/Step Audio EditX · 2 inputs · 3 parameters · 1 output

EditXSingleVoiceCloner

ComfyUI node: Voice cloning using loaded EditX models. Uses StepAudioTTS.clone(prompt_wav_path, prompt_text, target_text) Input: - models: dict from EditXModelLoader - prompt_audio_path: path to reference wav file - prompt_text: text spoken in the reference audio - target_text: new text to synthesize with cloned voice Output: - AUDIO: (waveform_tensor, sample_rate)

Vantage/Step Audio EditX · 2 inputs · 3 parameters · 1 output

EditXSingleVoiceClonerFromPath

ComfyUI node: Voice cloning using loaded EditX models. Uses StepAudioTTS.clone(prompt_wav_path, prompt_text, target_text) Input: - models: dict from EditXModelLoader - prompt_audio_path: path to reference wav file - prompt_text: text spoken in the reference audio - target_text: new text to synthesize with cloned voice Output: - AUDIO: (waveform_tensor, sample_rate)

Vantage/Step Audio EditX · 1 input · 4 parameters · 1 output

EditXSingleVoiceEditor

ComfyUI node: Audio editing using loaded EditX models. Uses StepAudioTTS.edit(input_audio_path, audio_text, edit_type, edit_info, text) Input: - models: dict from EditXModelLoader - input_audio_path: path to input wav file - audio_text: text corresponding to the input audio (can be empty for denoise/vad) - edit_type: one of the supported types ("emotion", "style", "vad", "denoise", "paralinguistic", "speed") - edit_info: subtype, e.g. "happy" / "sad" / "remove" etc. - target_text: used only for "paralinguistic" edits (new text) Output: - AUDIO: (waveform_tensor, sample_rate)

Vantage/Step Audio EditX · 2 inputs · 5 parameters · 1 output

EditXSingleVoiceEditorFromPath

ComfyUI node: Audio editing using loaded EditX models. Uses StepAudioTTS.edit(input_audio_path, audio_text, edit_type, edit_info, text) Input: - models: dict from EditXModelLoader - input_audio_path: path to input wav file - audio_text: text corresponding to the input audio (can be empty for denoise/vad) - edit_type: one of the supported types ("emotion", "style", "vad", "denoise", "paralinguistic", "speed") - edit_info: subtype, e.g. "happy" / "sad" / "remove" etc. - target_text: used only for "paralinguistic" edits (new text) Output: - AUDIO: (waveform_tensor, sample_rate)

Vantage/Step Audio EditX · 1 input · 6 parameters · 1 output

LoadSpeakers

Compact description unavailable for this node.

Vantage/Step Audio EditX · 6 inputs · 6 parameters · 1 output