# sampada-11/voice-deepfake-detector architecture

> Architecture diagram of the GitHub repository sampada-11/voice-deepfake-detector, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-10-06). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/sampada-11/voice-deepfake-detector
- Repository: https://github.com/sampada-11/voice-deepfake-detector

## Overview

Voice Deepfake Detector is a local Flask application backed by a classical acoustic-feature/SVM pipeline. A user uploads audio in the browser; Flask validates it, extracts 36 features, predicts REAL or FAKE with a confidence score, logs the result, and returns it for display. The repository also includes dataset inspection and construction, baseline and conversion-aware experiments, and final-model training. Source excerpts confirm the main inference and data-building calls; broader ML-stage links follow the README, while unsampled implementation details are not assumed.

Main entry points: [`app.py`](https://github.com/sampada-11/voice-deepfake-detector/blob/main/app.py) (Flask application), [`data`](https://github.com/sampada-11/voice-deepfake-detector/tree/main/data) (Feature dataset), [`ml/data_processing/build_dataset.py`](https://github.com/sampada-11/voice-deepfake-detector/blob/main/ml/data_processing/build_dataset.py) (Dataset builder), [`ml/train_final_model.py`](https://github.com/sampada-11/voice-deepfake-detector/blob/main/ml/train_final_model.py) (Final model training)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_web["Web detection"]
  node_browser["Browser UI<br/>[main.js]"]
  node_pages["Web pages"]
  node_server["Flask application<br/>[app.py]"]
  node_result["Prediction response"]
end

subgraph group_features["Acoustic features"]
  node_extract["Feature extraction"]
  node_vector["36-feature vector"]
end

subgraph group_data["Dataset preparation"]
  node_inspect["Dataset inspection<br/>[inspect_dataset.py]"]
  node_builder["Dataset builder<br/>[build_dataset.py]"]
  node_featuredata["Feature dataset"]
end

subgraph group_ml["Modeling and evaluation"]
  node_baseline["Baseline experiments<br/>[train.py]"]
  node_finaltrain["Final model training"]
  node_conversion["Conversion-aware evaluation"]
  node_savedmodel["Saved SVM pipeline<br/>[detector.pkl]"]
  node_metadata["Feature metadata"]
end

subgraph group_state["Runtime state"]
  node_history[("Detection history")]
end

node_user(("User"))
node_audio["Uploaded audio"]
node_deepvoice[("DeepVoice dataset")]

node_user -->|"selects audio"| node_browser
node_browser -->|"submits audio"| node_server
node_server -->|"extracts features"| node_extract
node_server -->|"loads and predicts"| node_savedmodel
node_extract -->|"returns vector"| node_vector
node_server -->|"writes and reads"| node_history
node_server -->|"returns prediction"| node_result
node_result -->|"displays result"| node_browser
node_user -->|"views pages"| node_pages
node_deepvoice -->|"inspects samples"| node_inspect
node_deepvoice -->|"builds from samples"| node_builder
node_builder -->|"extracts features"| node_extract
node_builder -->|"writes feature data"| node_featuredata
node_featuredata -->|"feeds experiments"| node_baseline
node_featuredata -->|"feeds training"| node_finaltrain
node_featuredata -->|"feeds evaluation"| node_conversion
node_finaltrain -->|"saves pipeline"| node_savedmodel
node_finaltrain -->|"saves feature order"| node_metadata
node_baseline -.->|"compares generalization"| node_conversion

click node_browser "https://github.com/sampada-11/voice-deepfake-detector/blob/main/static/js/main.js"
click node_pages "https://github.com/sampada-11/voice-deepfake-detector/tree/main/templates"
click node_server "https://github.com/sampada-11/voice-deepfake-detector/blob/main/app.py"
click node_extract "https://github.com/sampada-11/voice-deepfake-detector/blob/main/ml/data_processing/feature_extraction.py"
click node_inspect "https://github.com/sampada-11/voice-deepfake-detector/blob/main/ml/data_processing/inspect_dataset.py"
click node_builder "https://github.com/sampada-11/voice-deepfake-detector/blob/main/ml/data_processing/build_dataset.py"
click node_featuredata "https://github.com/sampada-11/voice-deepfake-detector/tree/main/data"
click node_baseline "https://github.com/sampada-11/voice-deepfake-detector/blob/main/ml/train.py"
click node_finaltrain "https://github.com/sampada-11/voice-deepfake-detector/blob/main/ml/train_final_model.py"
click node_conversion "https://github.com/sampada-11/voice-deepfake-detector/blob/main/ml/evaluation/conversion_aware_evaluation.py"
click node_savedmodel "https://github.com/sampada-11/voice-deepfake-detector/blob/main/models/detector.pkl"
click node_metadata "https://github.com/sampada-11/voice-deepfake-detector/blob/main/models/model_metadata.json"

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_browser,node_pages,node_server,node_result,node_user toneBlue
class node_extract,node_vector,node_deepvoice toneAmber
class node_inspect,node_builder,node_featuredata toneMint
class node_baseline,node_finaltrain,node_conversion,node_savedmodel,node_metadata toneRose
class node_history,node_audio toneIndigo
```

## Components

### Web detection

- [Browser UI](https://github.com/sampada-11/voice-deepfake-detector/blob/main/static/js/main.js): `static/js/main.js`
- [Web pages](https://github.com/sampada-11/voice-deepfake-detector/tree/main/templates): `templates`
- [Flask application](https://github.com/sampada-11/voice-deepfake-detector/blob/main/app.py): `app.py`
- **Prediction response**

### Acoustic features

- [Feature extraction](https://github.com/sampada-11/voice-deepfake-detector/blob/main/ml/data_processing/feature_extraction.py): `ml/data_processing/feature_extraction.py`
- **36-feature vector**

### Dataset preparation

- [Dataset inspection](https://github.com/sampada-11/voice-deepfake-detector/blob/main/ml/data_processing/inspect_dataset.py): `ml/data_processing/inspect_dataset.py`
- [Dataset builder](https://github.com/sampada-11/voice-deepfake-detector/blob/main/ml/data_processing/build_dataset.py): `ml/data_processing/build_dataset.py`
- [Feature dataset](https://github.com/sampada-11/voice-deepfake-detector/tree/main/data): `data`

### Modeling and evaluation

- [Baseline experiments](https://github.com/sampada-11/voice-deepfake-detector/blob/main/ml/train.py): `ml/train.py`
- [Final model training](https://github.com/sampada-11/voice-deepfake-detector/blob/main/ml/train_final_model.py): `ml/train_final_model.py`
- [Conversion-aware evaluation](https://github.com/sampada-11/voice-deepfake-detector/blob/main/ml/evaluation/conversion_aware_evaluation.py): `ml/evaluation/conversion_aware_evaluation.py`
- [Saved SVM pipeline](https://github.com/sampada-11/voice-deepfake-detector/blob/main/models/detector.pkl): `models/detector.pkl`
- [Feature metadata](https://github.com/sampada-11/voice-deepfake-detector/blob/main/models/model_metadata.json): `models/model_metadata.json`

### Runtime state

- **Detection history**

### Other components

- **User**
- **Uploaded audio**
- **DeepVoice dataset**

## Connections

- User → Browser UI: selects audio (evidence: `README.md`)
- Browser UI → Flask application: submits audio (evidence: `README.md`)
- Flask application → Feature extraction: extracts features (evidence: `app.py`)
- Flask application → Saved SVM pipeline: loads and predicts (evidence: `app.py`)
- Feature extraction → 36-feature vector: returns vector (evidence: `README.md`)
- Flask application → Detection history: writes and reads (evidence: `app.py`)
- Flask application → Prediction response: returns prediction (evidence: `README.md`)
- Prediction response → Browser UI: displays result (evidence: `README.md`)
- User → Web pages: views pages (evidence: `README.md`)
- DeepVoice dataset → Dataset inspection: inspects samples (evidence: `ml/data_processing/inspect_dataset.py`)
- DeepVoice dataset → Dataset builder: builds from samples (evidence: `ml/data_processing/build_dataset.py`)
- Dataset builder → Feature extraction: extracts features (evidence: `ml/data_processing/build_dataset.py`)
- Dataset builder → Feature dataset: writes feature data (evidence: `ml/data_processing/build_dataset.py`)
- Feature dataset → Baseline experiments: feeds experiments (evidence: `ml/train.py`)
- Feature dataset → Final model training: feeds training (evidence: `ml/train_final_model.py`)
- Feature dataset → Conversion-aware evaluation: feeds evaluation (evidence: `README.md`)
- Final model training → Saved SVM pipeline: saves pipeline (evidence: `README.md`)
- Final model training → Feature metadata: saves feature order (evidence: `README.md`)
- Baseline experiments → Conversion-aware evaluation: compares generalization (evidence: `README.md`)
