ComfyUI Node: YuE2 参考音色翻唱

YuE2 参考音色翻唱:

The YuE2ReferenceVoiceCover node is designed to facilitate the creation of vocal covers by leveraging a reference voice. This node allows you to transform an existing song audio into a new version that mimics the timbre and style of a specified reference voice. By adjusting various parameters, you can control the degree of similarity to the reference voice, the pitch, and other acoustic properties, making it a powerful tool for artists looking to experiment with different vocal styles or create unique cover versions of songs. The node's primary goal is to provide a seamless and intuitive way to generate high-quality vocal covers that maintain the essence of the original while introducing new vocal characteristics.

YuE2 参考音色翻唱 Input Parameters:

model

This parameter specifies the model to be used for the voice conversion process. It determines the underlying algorithm and data that will be applied to transform the song audio into a cover with the reference voice's characteristics.

song_audio

The song_audio parameter is the audio file of the song you wish to create a cover for. This input serves as the base audio that will be transformed using the reference voice.

reference_voice

This parameter is the audio file of the reference voice that you want the cover to emulate. It provides the vocal characteristics that will be applied to the song audio.

diffusion_steps

diffusion_steps controls the number of steps in the diffusion process, which affects the quality and detail of the voice conversion. It ranges from 4 to 50, with a default value of 30. More steps generally result in higher quality but require more processing time.

timbre_strength

This parameter adjusts the strength of the timbre effect applied from the reference voice. It ranges from 0.0 to 1.5, with a default value of 0.7. A higher value increases the influence of the reference voice's timbre on the final output.

auto_match_pitch

auto_match_pitch is a boolean parameter that, when enabled, automatically adjusts the pitch of the song audio to match the reference voice. This can help in achieving a more natural-sounding cover.

semitone_shift

The semitone_shift parameter allows you to shift the pitch of the song audio by a specified number of semitones, ranging from -12 to 12, with a default of 0. This can be used to fine-tune the pitch alignment between the song and the reference voice.

YuE2 参考音色翻唱 Output Parameters:

AUDIO

The AUDIO output is the transformed audio file that represents the song cover with the reference voice's characteristics applied. This is the primary output that you will use for further processing or playback.

YUE2_RESULT

This output provides metadata and additional information about the voice conversion process, including details about the job execution and any relevant parameters used.

STRING

The STRING output contains a JSON-formatted string with the status and results of the conversion process. This can be useful for debugging or logging purposes.

YuE2 参考音色翻唱 Usage Tips:

• Experiment with different timbre_strength values to find the right balance between the original song and the reference voice's characteristics.
• Use auto_match_pitch to ensure the cover sounds natural, especially when the original song and reference voice have different pitch ranges.
• Adjust semitone_shift to fine-tune the pitch if the automatic pitch matching does not yield the desired results.

YuE2 参考音色翻唱 Common Errors and Solutions:

"无法生成翻唱：转谱没有产生可用的 ABC；请提供核对后的 ABC"

• Explanation: This error occurs when the transcription process does not produce a valid ABC notation, which is necessary for generating the cover.
• Solution: Ensure that the transcription is correct and provide a verified ABC notation for the process to proceed.

"RVC 音色已被移动或删除，请刷新 ComfyUI 页面"

• Explanation: This error indicates that the reference voice has been moved or deleted from the system.
• Solution: Refresh the ComfyUI page to update the available voices and select a valid reference voice.

"该音色没有说话人 {sid}；可用说话人：{names or '无'}"

• Explanation: The selected reference voice does not have the specified speaker ID.
• Solution: Choose a speaker ID that is available for the selected reference voice.