sauim-detector 0.1.2

Bioacoustic processing and Pied tamarin (Saguinus bicolor) vocalization detector.

🐒 Sauim Detector

sauim-detector is a Python command-line tool for bioacoustic processing and automatic detection of Pied tamarin (Saguinus bicolor) vocalizations. It combines a pre-trained bird vocalization embedding model with a custom OCSVM classifier to detect the presence of tamarin calls in audio recordings. 🙈🙉🙊

Quick tutorial: Watch on YouTube

---

📦 Installation

Directly from pypi:

pip install sauim-detector==0.1.1

Directly from the GitHub repo (clone and install the package):

git clone https://github.com/juancolonna/Sauim.git
cd sauim-detector
pip install .

This will also install required dependencies:

• librosa
• numpy
• scipy
• tensorflow
• tensorflow-hub
• joblib
• soundfile

---

🚀 Usage

The CLI entry point is sauim-detector.

sauim-detector path/to/audio.wav

Options

• --save-audio, -s
  If set, saves the filtered signal as a .wav file alongside the labels.

---

📂 Outputs

1. Detection labels in Audacity label format:

   start_time    end_time    label
   0.00          7.20        sauim
   10.00         15.50       sauim
   20.00         30.80       sauim
   ....
   

   These can be imported directly into Audacity via
   File → Import → Labels….

2. Filtered audio (optional, if --save-audio is used):
   A .wav file containing the processed signal.
   Example: audio_filtered.wav

---

📝 Example

# Run classification
sauim-detector recordings/example.wav

# Run classification and also save filtered audio
sauim-detector recordings/example.wav --save-audio

Output:

Total detections: 2
✅ Labels saved as: recordings/example_detections.txt
✅ Filtered signal saved as: recordings/example_filtered.wav

---

⚠️ Notes

• Input files must be in .wav format.
• The default sampling rate is 32 kHz. Files will be resampled automatically if needed.
• Labels are generated based on classifier decisions (OCSVM) and they need manual validation on Audacity.