How freyzamarshall02/aicovergen-mod works
AICoverGen turns a YouTube link or local audio file into an AI voice cover. Users can run the generation workflow through the WebUI or CLI; the main pipeline separates vocals and instrumentals, converts the vocals with an RVC voice model, then mixes the result and produces audio output. The WebUI also supports managing voice models. Internal RVC inference uses pitch extraction and a synthesizer with selectable vocoder implementations. Several relationships are supported only by the README or SOURCE INDEX; unsampled implementation details are not inferred.
Main entry points: synthesizers.py RVC synthesizer · webui.py Web UI · main.py Cover pipeline · rvc_module.py RVC wrapper
Components 26
Cover generation
- Cover pipeline
src/main.py - Vocal separation
src/mdx.py - Audio mixing
src/main.py
RVC conversion
- RVC wrapper
src/rvc_module.py - Conversion pipeline
src/rvc/infer/pipeline.py - Pitch extraction
src/rvc/lib/predictors/f0.py - Pitch models
src/rvc/lib/predictors/RMVPE.py - RVC synthesizer
src/rvc/lib/algorithm/synthesizers.py - Text encoder
src/rvc/lib/algorithm/encoders.py - Attention layers
src/rvc/lib/algorithm/attentions.py - Algorithm utilities
src/rvc/lib/algorithm/commons.py - Residual blocks
src/rvc/lib/algorithm/residuals.py - WaveNet modules
src/rvc/lib/algorithm/modules.py - HiFi-GAN vocoders
src/rvc/lib/algorithm/generators/hifigan.py - MRF vocoder
src/rvc/lib/algorithm/generators/hifigan_mrf.py - NSF vocoder
src/rvc/lib/algorithm/generators/hifigan_nsf.py - RefineGAN vocoder
src/rvc/lib/algorithm/generators/refinegan.py
Model management
- Voice models
rvc_models - Model download
src/download_models.py - Separation models
mdxnet_models
Other components
- User
- Song source
- Cover audio
- Model hosting
Connections 28
- User → Web UI: uses
README.md - User → CLI entry: runs
README.md - User → Song source: provides
README.md - Web UI → Cover pipeline: starts generation
README.md - CLI entry → Cover pipeline: runs
README.md - Cover pipeline → Vocal separation: separates stems
src/main.py - Cover pipeline → RVC wrapper: converts vocals
src/main.py - Cover pipeline → Audio mixing: mixes audio
src/main.py - Audio mixing → Cover audio: writes
README.md - RVC wrapper → Conversion pipeline: calls
src/rvc_module.py - RVC wrapper → RVC synthesizer: loads
src/rvc_module.py - RVC wrapper → Voice models: loads model
- Conversion pipeline → Pitch extraction: requests pitch
src/rvc/infer/pipeline.py - Pitch extraction → Pitch models: uses
src/rvc/lib/predictors/f0.py - RVC synthesizer → Text encoder: uses
src/rvc/lib/algorithm/synthesizers.py - RVC synthesizer → HiFi-GAN vocoders: selects vocoder
src/rvc/lib/algorithm/synthesizers.py - RVC synthesizer → MRF vocoder: selects vocoder
src/rvc/lib/algorithm/synthesizers.py - RVC synthesizer → NSF vocoder: selects vocoder
src/rvc/lib/algorithm/synthesizers.py - RVC synthesizer → RefineGAN vocoder: selects vocoder
src/rvc/lib/algorithm/synthesizers.py - Text encoder → Attention layers: uses
src/rvc/lib/algorithm/encoders.py - Attention layers → Algorithm utilities: uses
src/rvc/lib/algorithm/attentions.py - HiFi-GAN vocoders → Residual blocks: uses
src/rvc/lib/algorithm/generators/hifigan.py - Residual blocks → WaveNet modules: uses
src/rvc/lib/algorithm/residuals.py - Residual blocks → Algorithm utilities: uses
src/rvc/lib/algorithm/residuals.py - Web UI → Voice models: manages models
README.md - Web UI → Model download: downloads via
src/webui.py - Model download → Model hosting: downloads from
- Web UI → Separation models: fetches base models