USVSEG: syllable segmentator for rodents' ultrasonic vocalization

We provide USVSEG (developed by Dr. Tachibana), GUI-based MATLAB script for robust segmentation of rodents' ultrasonic vocalizations. This system is useful for observation of social communication such as male-female interaction because it is very robust for background noise. USVSEG automatically detects syllables and output the data of start and end time of each syllable, duration, peak frequency, and peak amplitude (-dB). Please see output sample data below. Each syllable can be exported as a jpeg image and/or sound file (wav). NO toolbox for MATLAB is needed.


とにかくノイズに強いのが特徴です。なので、pupUSVsのみならず、雄-雌間や雌-雌間などノイズの多い環境での録音データもなんなく処理してくれます。検出された音節のdurationやinterval、Frequencyや-dBを算出します。その音節をjpeg画像 and/or wav音源(もしくは検出したシグナルの行列データ)としてoutputしてくれます。



How to Use At a glance!

An example of output data

Data of a wav file are exported as csv file (although an Excel file is embedded in the right panel here because of google site format).


Video 1

How to decide optimal settings with short sounds

Video 2

How to apply and analyze a long sound file

If you put "multiple files" button, more than one files in a same folder can be analyzed at once (processes shown in video 2 are automatically repeated for each file).

Download & Information (now published in PLoS ONE)

(Please find the MATLAB file and user's manual)

Parts of this project was supported by Grant-in Scientific Research on Innovation Areas "Integrative Research toward Elucidation of Generative Brain Systems for Individuality" and "Intergrative Studies of Language Evolution for Co-creative Communication".

USVSEGの開発と精度評価は、文部科学省 科学研究費補助金 新学術領域「個性」創発脳共創言語進化の領域間共同研究の一環として行われました。

Other software already published in the world is presented on mouseTube website (menu -> community).