🎧 SpecView Local Audio Browser

Browse. Inspect. Tag. All without leaving your machine.

I built this because I kept running into the same problem: a large folder of audio files, a need to spot-check a handful of them, and absolutely no desire to upload 40GB to a web service just to look at three spectrograms.

SpecView Local is a lightweight, local-first audio browser. Drag in files or folders, navigate a tree view, render spectrograms only for what you're actively viewing, and optionally run CED-Tiny audio tagging on the current file — on demand, not in bulk.

Having written my own audio visualization tools before, I took what I liked from SpecView's interaction design and rebuilt it with my own UI framework and color scheme.

---

💡 Why it's different

Most audio tools want to ingest your whole dataset upfront. This one doesn't. Files stay on your disk; the app only reads what you're looking at right now. That makes it fast to start, easy on memory, and well-suited for the kind of open-ended, exploratory inspection that doesn't fit neatly into a pipeline.

---

✨ Highlights

🗂️ Lazy local workflow Files are referenced, not uploaded. Only the currently viewed file is ever read or rendered.

📊 On-demand spectrogram rendering Click a file, get a spectrogram — no batch pre-rendering. Fully adjustable parameters:

• Transform: log-mel, log-spec, mel, spec
• sample_rate, n_fft, hop_length, n_mels, fmin, fmax
• Colormap options: magma, inferno, viridis, plasma, cividis

📁 Built for large collections Folder tree navigation and automatic grouping of same-name files with different extensions keep things manageable even when you have thousands of files.

🏷️ Audio tagging with CED-Tiny Click Analyze (or press Enter) on any file to run lightweight audio classification and get Top-K labels with confidence scores. Nothing runs in the background unless you ask for it.

🌐 English / Chinese UI Switch between English and Chinese at any time with the 中 / EN toggle in the toolbar — preference is saved across sessions.

⌨️ Keyboard-first navigation Every common action has a shortcut. Press ? anytime to see the full list.

---

🚀 Getting Started

1. Clone the repo

git clone https://github.com/Abusir/SpecViewLocal.git
cd SpecViewLocal

2. Create a conda environment

conda create -n specview python=3.10 -y
conda activate specview

3. Install dependencies

python -m pip install --upgrade pip
python -m pip install -r requirements.txt

If you run into ONNX issues with the analysis feature:

python -m pip install -U "optimum-onnx[onnxruntime]"

4. Run

python app.py --host 127.0.0.1 --port 8000

Open http://127.0.0.1:8000 and you're in.

---

🎮 How to Use

📥 Import audio

Drag files or a folder into the left panel, or use the file/folder picker. The app keeps local references only — no full-folder upload.

🔍 Browse

Click any file in the tree to load it. Keyboard shortcuts:

📈 Inspect spectrograms

Click a file to render its spectrogram. Open the Settings panel to adjust rendering parameters. Click anywhere on the spectrogram to jump to that playback position.

🤖 Run audio analysis

Click Analyze on the current file to run CED-Tiny and see Top-K audio tags. The first run may download the model from Hugging Face — subsequent runs are fast.

---

🗃️ Project Structure

.
├── app.py
├── requirements.txt
├── templates/
│   └── index.html
└── static/
    ├── app.js
    └── styles.css

---

🙏 Acknowledgements

The interaction design and spectrogram browser concept are heavily inspired by SpecView by RicherMans. If you like this tool, go star the original.

Audio analysis is powered by CED-Tiny, a compact audio tagging model from the CED family. Impressively capable for its size.

---

📄 License

MIT — do whatever you want with it, just keep the attribution.