Batch Audio Vocal Separation Tool

Batch separate vocals and accompaniment from music, with support for recording noise reduction, echo and reverb removal, vocal extraction, and local offline processing.

Bug report

Loading tool, please wait...

If the online tool fails to load or run, try the desktop tool.https://tools.yikeaigc.com/

Tool Usage

Return to old version

A desktop version is available with more powerful, stable features. Open desktop version

Instructions

Software Usage Instructions

Add audio files: Click "Add File" or "Add Folder" to import the audio you want to process. Batch import is supported, and you can check "Recursive Subdirectories" to automatically scan all audio files in the folder. Common formats such as MP3, WAV, FLAC, M4A, AAC, and OGG are supported, and audio tracks in videos such as MP4 and MKV can also be processed directly.
Select a processing mode:
- Music vocal separation: Separate a song into two tracks: vocals and accompaniment. Different models such as high quality, recommended, and fast are available.
- Recording noise reduction and vocal extraction: Extract clear vocals from materials with background noise, such as recordings, podcasts, and live streams.
- Echo and reverb removal: Remove echo and reverb from vocals, and it can be processed in sequence with noise reduction mode.
Set the model directory: When using it for the first time, place the model files in the models folder at the same level as the software, and the software will automatically recognize them. If a model is missing, you can obtain the corresponding model file from the download link as prompted.
Configure processing options:
- Output format: WAV, FLAC, and MP3 are available.
- Parameter presets: Presets such as balanced default, speed priority, quality priority, and low memory are provided.
- Output directory: Specify where to save the separation results, and optionally preserve the original directory structure.
Start processing: Click the "Start Processing" button, and the software will process all audio files in the queue one by one. Progress and logs will be displayed in real time.
View results: After processing is complete, you can find the separated vocal and accompaniment files in the output directory. The file names will indicate the corresponding track type.

FAQ

A: Music vocal separation is designed for songs, splitting singing vocals and instrumental accompaniment into two tracks. It is suitable for creating accompaniments, covers, and karaoke. Recording noise reduction and vocal extraction is designed for recordings, podcasts, live streams, and similar materials, focusing on removing background noise while preserving spoken or sung vocals. The two modes are intended for different scenarios and can be selected according to the material type.

A: Common audio formats such as MP3, WAV, FLAC, M4A, AAC, OGG, OPUS, WMA, APE, and AIFF are supported. Audio tracks in video files such as MP4, MKV, MOV, AVI, and WEBM can also be read directly for separation. Output can be selected in WAV, FLAC, or MP3 format.

A: Vocal separation relies on AI model files. Please place the corresponding model files (such as Kim_Vocal_2.onnx, UVR-DeNoise.pth, etc.) in the models folder at the same level as the software, then select a mode. Model download links are provided in the software; download the files by name and place them in the directory, and they will be automatically recognized.

A: No. All separation and noise reduction processing is completed locally on your device. Audio files will not be uploaded to the server, and the processed results are saved directly to the local output directory, helping protect your material privacy and data security.

A: AI vocal separation requires a certain amount of CPU and memory resources. Recommendations: 1) select the "Speed First" or "Low Memory" parameter preset; 2) process large numbers of files in batches; 3) close other programs that use a lot of memory; 4) if you have an NVIDIA graphics card and the required runtime environment installed, you can select GPU mode to improve speed.

A: Yes. The software supports two-stage processing: first use a noise reduction model to remove background noise and extract vocals, then use an echo removal and dereverberation model for further processing. Chaining the two steps makes vocals clearer and cleaner, making it suitable for post-production cleanup of recordings, podcasts, and livestream materials.

A: After processing is complete, the software will generate the corresponding files in the output directory, and the file names will indicate the track type (for example, Vocals means vocals, and Instrumental means accompaniment). You can quickly find the vocals or accompaniment track you need based on the file name suffix.