# charantejagoud2006/audio_inpainting architecture

> Architecture diagram of the GitHub repository charantejagoud2006/audio_inpainting, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-10-28). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/charantejagoud2006/audio_inpainting
- Repository: https://github.com/charantejagoud2006/audio_inpainting

## Overview

DPAI is a research audio-inpainting application: users provide audio and a gap mask (or request synthetic gaps), the system reconstructs missing audio using a deep-prior-trained MultiResUNet, then saves artifacts and metrics. The graph also preserves the benchmark/summarization and interactive web-demo capabilities documented in the README. Relationships reflect the excerpts and SOURCE INDEX; unsampled benchmark internals and some web UI wiring are not expanded.

Main entry points: [`src/pipeline/inpainter.py`](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/pipeline/inpainter.py) (Inpainting pipeline), [`app.py`](https://github.com/charantejagoud2006/audio_inpainting/blob/main/app.py) (CLI commands), [`src/web/server.py`](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/web/server.py) (Web API and jobs), [`src/losses/total_loss.py`](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/losses/total_loss.py) (Combined loss)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_entry["Application entry"]
  node_app["CLI commands<br/>[app.py]"]
  node_benchmark["Benchmark suite<br/>[benchmark.py]"]
end

subgraph group_audio["Audio preparation"]
  node_loader["Audio loading<br/>[loader.py]"]
  node_corruption["Gap masks<br/>[corruption.py]"]
  node_stft["STFT and masks<br/>[stft.py]"]
end

subgraph group_reconstruction["Reconstruction"]
  node_inpainter["Inpainting pipeline<br/>[inpainter.py]"]
  node_deepprior["Deep-prior fitting<br/>[deep_prior.py]"]
  node_noise["Input noise<br/>[noise.py]"]
  node_multires["MultiResUNet<br/>[multires_unet.py]"]
  node_harmonic["Harmonic convolution<br/>[harmonic_conv.py]"]
  node_total_loss["Combined loss<br/>[total_loss.py]"]
  node_masked_mse["Masked MSE<br/>[masked_mse.py]"]
  node_multiscale_loss["Multiscale loss<br/>[multiscale_loss.py]"]
end

subgraph group_reporting["Evaluation and artifacts"]
  node_metrics["Run metrics<br/>[metrics.py]"]
  node_evaluation["NMSE and PESQ<br/>[pesq.py]"]
  node_run_io["Run artifacts<br/>[io.py]"]
  node_visualization["Plots and previews<br/>[visualization.py]"]
end

subgraph group_web["Web demo"]
  node_server["Web API and jobs<br/>[server.py]"]
  node_web_ui["Interactive interface<br/>[app.js]"]
end

node_user(("CLI user"))
node_browser(("Browser user"))

node_user -->|"runs commands"| node_app
node_app -->|"loads audio"| node_loader
node_app -->|"creates gaps"| node_corruption
node_app -->|"starts inpainting"| node_inpainter
node_app -->|"runs benchmark"| node_benchmark
node_app -->|"saves run"| node_run_io
node_inpainter -->|"transforms audio"| node_stft
node_inpainter -->|"builds model"| node_multires
node_inpainter -->|"optimizes model"| node_deepprior
node_inpainter -->|"creates input"| node_noise
node_inpainter -->|"configures loss"| node_total_loss
node_deepprior -->|"perturbs input"| node_noise
node_deepprior -->|"uses default loss"| node_masked_mse
node_multires -->|"uses layers"| node_harmonic
node_total_loss -->|"combines term"| node_masked_mse
node_total_loss -->|"combines term"| node_multiscale_loss
node_total_loss -->|"converts spectrogram"| node_stft
node_multiscale_loss -->|"computes spectrograms"| node_stft
node_run_io -->|"writes plots"| node_visualization
node_run_io -->|"records metrics"| node_metrics
node_metrics -->|"computes scores"| node_evaluation
node_browser -->|"uses demo"| node_server
node_server -->|"loads uploads"| node_loader
node_server -->|"creates gaps"| node_corruption
node_server -->|"queues inpainting"| node_inpainter
node_server -->|"saves runs"| node_run_io
node_web_ui -.->|"calls API"| node_server

click node_app "https://github.com/charantejagoud2006/audio_inpainting/blob/main/app.py"
click node_benchmark "https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/audio/benchmark.py"
click node_loader "https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/audio/loader.py"
click node_corruption "https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/audio/corruption.py"
click node_stft "https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/audio/stft.py"
click node_inpainter "https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/pipeline/inpainter.py"
click node_deepprior "https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/training/deep_prior.py"
click node_noise "https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/training/noise.py"
click node_multires "https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/models/multires_unet.py"
click node_harmonic "https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/models/harmonic_conv.py"
click node_total_loss "https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/losses/total_loss.py"
click node_masked_mse "https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/losses/masked_mse.py"
click node_multiscale_loss "https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/losses/multiscale_loss.py"
click node_metrics "https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/pipeline/metrics.py"
click node_evaluation "https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/evaluation/pesq.py"
click node_run_io "https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/pipeline/io.py"
click node_visualization "https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/utils/visualization.py"
click node_server "https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/web/server.py"
click node_web_ui "https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/web/static/app.js"

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_app,node_benchmark,node_user,node_browser toneBlue
class node_loader,node_corruption,node_stft toneAmber
class node_inpainter,node_deepprior,node_noise,node_multires,node_harmonic,node_total_loss,node_masked_mse,node_multiscale_loss toneMint
class node_metrics,node_evaluation,node_run_io,node_visualization toneRose
class node_server,node_web_ui toneIndigo
```

## Components

### Application entry

- [CLI commands](https://github.com/charantejagoud2006/audio_inpainting/blob/main/app.py): `app.py`
- [Benchmark suite](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/audio/benchmark.py): `src/audio/benchmark.py`

### Audio preparation

- [Audio loading](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/audio/loader.py): `src/audio/loader.py`
- [Gap masks](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/audio/corruption.py): `src/audio/corruption.py`
- [STFT and masks](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/audio/stft.py): `src/audio/stft.py`

### Reconstruction

- [Inpainting pipeline](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/pipeline/inpainter.py): `src/pipeline/inpainter.py`
- [Deep-prior fitting](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/training/deep_prior.py): `src/training/deep_prior.py`
- [Input noise](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/training/noise.py): `src/training/noise.py`
- [MultiResUNet](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/models/multires_unet.py): `src/models/multires_unet.py`
- [Harmonic convolution](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/models/harmonic_conv.py): `src/models/harmonic_conv.py`
- [Combined loss](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/losses/total_loss.py): `src/losses/total_loss.py`
- [Masked MSE](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/losses/masked_mse.py): `src/losses/masked_mse.py`
- [Multiscale loss](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/losses/multiscale_loss.py): `src/losses/multiscale_loss.py`

### Evaluation and artifacts

- [Run metrics](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/pipeline/metrics.py): `src/pipeline/metrics.py`
- [NMSE and PESQ](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/evaluation/pesq.py): `src/evaluation/pesq.py`
- [Run artifacts](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/pipeline/io.py): `src/pipeline/io.py`
- [Plots and previews](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/utils/visualization.py): `src/utils/visualization.py`

### Web demo

- [Web API and jobs](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/web/server.py): `src/web/server.py`
- [Interactive interface](https://github.com/charantejagoud2006/audio_inpainting/blob/main/src/web/static/app.js): `src/web/static/app.js`

### Other components

- **CLI user**
- **Browser user**

## Connections

- CLI user → CLI commands: runs commands (evidence: `README.md`)
- CLI commands → Audio loading: loads audio (evidence: `app.py`)
- CLI commands → Gap masks: creates gaps (evidence: `app.py`)
- CLI commands → Inpainting pipeline: starts inpainting (evidence: `app.py`)
- CLI commands → Benchmark suite: runs benchmark (evidence: `app.py`)
- CLI commands → Run artifacts: saves run (evidence: `app.py`)
- Inpainting pipeline → STFT and masks: transforms audio (evidence: `src/pipeline/inpainter.py`)
- Inpainting pipeline → MultiResUNet: builds model (evidence: `src/pipeline/inpainter.py`)
- Inpainting pipeline → Deep-prior fitting: optimizes model (evidence: `src/pipeline/inpainter.py`)
- Inpainting pipeline → Input noise: creates input (evidence: `src/pipeline/inpainter.py`)
- Inpainting pipeline → Combined loss: configures loss (evidence: `src/pipeline/inpainter.py`)
- Deep-prior fitting → Input noise: perturbs input (evidence: `src/training/deep_prior.py`)
- Deep-prior fitting → Masked MSE: uses default loss (evidence: `src/training/deep_prior.py`)
- MultiResUNet → Harmonic convolution: uses layers (evidence: `src/models/multires_unet.py`)
- Combined loss → Masked MSE: combines term (evidence: `src/losses/total_loss.py`)
- Combined loss → Multiscale loss: combines term (evidence: `src/losses/total_loss.py`)
- Combined loss → STFT and masks: converts spectrogram (evidence: `src/losses/total_loss.py`)
- Multiscale loss → STFT and masks: computes spectrograms (evidence: `src/losses/multiscale_loss.py`)
- Run artifacts → Plots and previews: writes plots (evidence: `src/pipeline/io.py`)
- Run artifacts → Run metrics: records metrics (evidence: `src/pipeline/io.py`)
- Run metrics → NMSE and PESQ: computes scores (evidence: `src/pipeline/metrics.py`)
- Browser user → Web API and jobs: uses demo (evidence: `README.md`)
- Web API and jobs → Audio loading: loads uploads (evidence: `src/web/server.py`)
- Web API and jobs → Gap masks: creates gaps (evidence: `src/web/server.py`)
- Web API and jobs → Inpainting pipeline: queues inpainting (evidence: `src/web/server.py`)
- Web API and jobs → Run artifacts: saves runs (evidence: `src/web/server.py`)
- Interactive interface → Web API and jobs: calls API
