# windywhu/speech-digit-recognition architecture

> Architecture diagram of the GitHub repository windywhu/speech-digit-recognition, 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/windywhu/speech-digit-recognition
- Repository: https://github.com/windywhu/speech-digit-recognition

## Overview

This is a Python desktop application for recognizing Mandarin digits zero through nine, distributed with pretrained CNN and RBF-SVM models. The main user flow is WAV or microphone input into the desktop window, model-specific preprocessing and inference, then displayed predictions, charts, playback, and exportable history. A separate research workflow prepares speaker-isolated datasets, trains both model families, and evaluates split, robustness, and external-speaker performance. The GUI excerpt is partial, so its detailed chart and history internals are represented at the documented capability level.

Main entry points: [`speech-digit-recognition/语音识别窗口.py`](https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/%E8%AF%AD%E9%9F%B3%E8%AF%86%E5%88%AB%E7%AA%97%E5%8F%A3.py) (Recognition window), [`speech-digit-recognition/CNN方法/outputs/cnn_best_model.pt`](https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/CNN%E6%96%B9%E6%B3%95/outputs/cnn_best_model.pt) (Saved CNN model), [`speech-digit-recognition/传统方法/outputs/best_model.joblib`](https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/%E4%BC%A0%E7%BB%9F%E6%96%B9%E6%B3%95/outputs/best_model.joblib) (Saved SVM model), [`speech-digit-recognition/data`](https://github.com/windywhu/speech-digit-recognition/tree/main/speech-digit-recognition/data) (Local WAV dataset)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_desktop["Desktop recognition"]
  node_gui["Recognition window<br/>[语音识别窗口.py]"]
  node_audio_in["Audio input and preprocessing<br/>[语音识别窗口.py]"]
  node_visuals["Charts and history<br/>[语音识别窗口.py]"]
end

subgraph group_models["Recognition models"]
  node_cnn_infer["CNN probabilities<br/>[predict_probs.py]"]
  node_svm_infer["RBF-SVM prediction<br/>[run_experiment.py]"]
  node_cnn_train["CNN training<br/>[train_cnn.py]"]
  node_svm_train["SVM training<br/>[run_experiment.py]"]
  node_cnn_model["Saved CNN model<br/>[cnn_best_model.pt]"]
  node_svm_model["Saved SVM model<br/>[best_model.joblib]"]
end

subgraph group_research["Training and evaluation"]
  node_cnn_eval["CNN split evaluation<br/>[evaluate_splits.py]"]
  node_svm_eval["SVM split evaluation<br/>[evaluate_splits.py]"]
  node_preprocess_eval["Preprocessing analysis"]
  node_robust_eval["Robustness evaluation"]
  node_external_eval["External-speaker evaluation"]
end

subgraph group_data["Dataset preparation"]
  node_split["Speaker-isolated split"]
  node_local_data[("Local WAV dataset")]
end

node_user(("User"))
node_audio["WAV / microphone"]

node_user -->|"uses"| node_gui
node_audio -->|"provides audio"| node_gui
node_gui -->|"loads and records"| node_audio_in
node_gui -->|"requests CNN inference"| node_cnn_infer
node_gui -->|"requests SVM inference"| node_svm_infer
node_gui -->|"presents results"| node_visuals
node_cnn_infer -->|"loads weights"| node_cnn_model
node_svm_infer -.->|"loads model"| node_svm_model
node_local_data -->|"supplies recordings"| node_split
node_split -->|"provides split"| node_cnn_train
node_split -->|"provides split"| node_svm_train
node_cnn_train -->|"writes checkpoint"| node_cnn_model
node_svm_train -->|"writes model"| node_svm_model
node_cnn_train -->|"provides checkpoint and cache"| node_cnn_eval
node_svm_train -->|"provides model and features"| node_svm_eval
node_local_data -->|"supplies WAV files"| node_preprocess_eval
node_local_data -->|"supplies held-out audio"| node_robust_eval
node_cnn_model -->|"evaluated by"| node_robust_eval
node_svm_model -->|"evaluated by"| node_robust_eval
node_cnn_model -->|"evaluated by"| node_external_eval
node_svm_model -->|"evaluated by"| node_external_eval

click node_gui "https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/%E8%AF%AD%E9%9F%B3%E8%AF%86%E5%88%AB%E7%AA%97%E5%8F%A3.py"
click node_audio_in "https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/%E8%AF%AD%E9%9F%B3%E8%AF%86%E5%88%AB%E7%AA%97%E5%8F%A3.py"
click node_visuals "https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/%E8%AF%AD%E9%9F%B3%E8%AF%86%E5%88%AB%E7%AA%97%E5%8F%A3.py"
click node_cnn_infer "https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/CNN%E6%96%B9%E6%B3%95/predict_probs.py"
click node_svm_infer "https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/%E4%BC%A0%E7%BB%9F%E6%96%B9%E6%B3%95/run_experiment.py"
click node_cnn_train "https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/CNN%E6%96%B9%E6%B3%95/train_cnn.py"
click node_svm_train "https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/%E4%BC%A0%E7%BB%9F%E6%96%B9%E6%B3%95/run_experiment.py"
click node_split "https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/prepare_dataset_split.py"
click node_local_data "https://github.com/windywhu/speech-digit-recognition/tree/main/speech-digit-recognition/data"
click node_cnn_eval "https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/CNN%E6%96%B9%E6%B3%95/evaluate_splits.py"
click node_svm_eval "https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/%E4%BC%A0%E7%BB%9F%E6%96%B9%E6%B3%95/evaluate_splits.py"
click node_preprocess_eval "https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/%E4%BC%A0%E7%BB%9F%E6%96%B9%E6%B3%95/preprocess_stats.py"
click node_robust_eval "https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/evaluate_robustness.py"
click node_external_eval "https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/evaluate_external_speakers.py"
click node_cnn_model "https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/CNN%E6%96%B9%E6%B3%95/outputs/cnn_best_model.pt"
click node_svm_model "https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/%E4%BC%A0%E7%BB%9F%E6%96%B9%E6%B3%95/outputs/best_model.joblib"

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_gui,node_audio_in,node_visuals,node_user toneBlue
class node_cnn_infer,node_svm_infer,node_cnn_train,node_svm_train,node_cnn_model,node_svm_model toneAmber
class node_cnn_eval,node_svm_eval,node_preprocess_eval,node_robust_eval,node_external_eval toneMint
class node_split,node_local_data toneRose
class node_audio toneIndigo
```

## Components

### Desktop recognition

- [Recognition window](https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/%E8%AF%AD%E9%9F%B3%E8%AF%86%E5%88%AB%E7%AA%97%E5%8F%A3.py): `speech-digit-recognition/语音识别窗口.py`
- [Audio input and preprocessing](https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/%E8%AF%AD%E9%9F%B3%E8%AF%86%E5%88%AB%E7%AA%97%E5%8F%A3.py): `speech-digit-recognition/语音识别窗口.py`
- [Charts and history](https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/%E8%AF%AD%E9%9F%B3%E8%AF%86%E5%88%AB%E7%AA%97%E5%8F%A3.py): `speech-digit-recognition/语音识别窗口.py`

### Recognition models

- [CNN probabilities](https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/CNN%E6%96%B9%E6%B3%95/predict_probs.py): `speech-digit-recognition/CNN方法/predict_probs.py`
- [RBF-SVM prediction](https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/%E4%BC%A0%E7%BB%9F%E6%96%B9%E6%B3%95/run_experiment.py): `speech-digit-recognition/传统方法/run_experiment.py`
- [CNN training](https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/CNN%E6%96%B9%E6%B3%95/train_cnn.py): `speech-digit-recognition/CNN方法/train_cnn.py`
- [SVM training](https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/%E4%BC%A0%E7%BB%9F%E6%96%B9%E6%B3%95/run_experiment.py): `speech-digit-recognition/传统方法/run_experiment.py`
- [Saved CNN model](https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/CNN%E6%96%B9%E6%B3%95/outputs/cnn_best_model.pt): `speech-digit-recognition/CNN方法/outputs/cnn_best_model.pt`
- [Saved SVM model](https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/%E4%BC%A0%E7%BB%9F%E6%96%B9%E6%B3%95/outputs/best_model.joblib): `speech-digit-recognition/传统方法/outputs/best_model.joblib`

### Training and evaluation

- [CNN split evaluation](https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/CNN%E6%96%B9%E6%B3%95/evaluate_splits.py): `speech-digit-recognition/CNN方法/evaluate_splits.py`
- [SVM split evaluation](https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/%E4%BC%A0%E7%BB%9F%E6%96%B9%E6%B3%95/evaluate_splits.py): `speech-digit-recognition/传统方法/evaluate_splits.py`
- [Preprocessing analysis](https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/%E4%BC%A0%E7%BB%9F%E6%96%B9%E6%B3%95/preprocess_stats.py): `speech-digit-recognition/传统方法/preprocess_stats.py`
- [Robustness evaluation](https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/evaluate_robustness.py): `speech-digit-recognition/evaluate_robustness.py`
- [External-speaker evaluation](https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/evaluate_external_speakers.py): `speech-digit-recognition/evaluate_external_speakers.py`

### Dataset preparation

- [Speaker-isolated split](https://github.com/windywhu/speech-digit-recognition/blob/main/speech-digit-recognition/prepare_dataset_split.py): `speech-digit-recognition/prepare_dataset_split.py`
- [Local WAV dataset](https://github.com/windywhu/speech-digit-recognition/tree/main/speech-digit-recognition/data): `speech-digit-recognition/data`

### Other components

- **User**
- **WAV / microphone**

## Connections

- User → Recognition window: uses (evidence: `README.md`)
- WAV / microphone → Recognition window: provides audio (evidence: `README.md`)
- Recognition window → Audio input and preprocessing: loads and records (evidence: `README.md`)
- Recognition window → CNN probabilities: requests CNN inference (evidence: `speech-digit-recognition/语音识别窗口.py`)
- Recognition window → RBF-SVM prediction: requests SVM inference (evidence: `speech-digit-recognition/语音识别窗口.py`)
- Recognition window → Charts and history: presents results (evidence: `README.md`)
- CNN probabilities → Saved CNN model: loads weights (evidence: `speech-digit-recognition/CNN方法/predict_probs.py`)
- RBF-SVM prediction → Saved SVM model: loads model
- Local WAV dataset → Speaker-isolated split: supplies recordings (evidence: `README.md`)
- Speaker-isolated split → CNN training: provides split (evidence: `README.md`)
- Speaker-isolated split → SVM training: provides split (evidence: `README.md`)
- CNN training → Saved CNN model: writes checkpoint (evidence: `README.md`)
- SVM training → Saved SVM model: writes model (evidence: `README.md`)
- CNN training → CNN split evaluation: provides checkpoint and cache (evidence: `README.md`)
- SVM training → SVM split evaluation: provides model and features (evidence: `README.md`)
- Local WAV dataset → Preprocessing analysis: supplies WAV files (evidence: `speech-digit-recognition/传统方法/preprocess_stats.py`)
- Local WAV dataset → Robustness evaluation: supplies held-out audio (evidence: `speech-digit-recognition/evaluate_robustness.py`)
- Saved CNN model → Robustness evaluation: evaluated by (evidence: `speech-digit-recognition/evaluate_robustness.py`)
- Saved SVM model → Robustness evaluation: evaluated by (evidence: `speech-digit-recognition/evaluate_robustness.py`)
- Saved CNN model → External-speaker evaluation: evaluated by (evidence: `speech-digit-recognition/evaluate_external_speakers.py`)
- Saved SVM model → External-speaker evaluation: evaluated by (evidence: `speech-digit-recognition/evaluate_external_speakers.py`)
