Gecko

Gecko allows efficient and effective segmentation of the voice signal by speaker as well as annotation of the linguistic content of the conversation. A key feature of Gecko is the presentation of the output of automatic segmentation and transcription systems in an intuitive user interface for editing. Gecko allows annotation of Voice Activity Detection (VAD), Diarization, Speaker Identiļ¬cation and ASR outputs on a large scale, and leads to faster and more accurate annotated datasets.

Features

  • Supports the annotating process of different stages of a conversation: voice detection, diarization, identification and transcription.
  • Provides an efficient and convenient tool for annotating audio files.
  • Visualize the annotation of several different sources at once.
  • Refine existing annotation files
  • Compare different annotating files to find discrepancies between different systems or annotators.
  • No server side is needed - easy installation.
  • Supports different formats such as RTTM, CTM, JSON, CSV.
  • Increased productivity using keyboard shortcuts

Technological Stack

Gecko is written in Javascript and is based on Angular.js V1.X.
The audio player uses the popular wavesurfer.js library.

Deployment and Installation

See this page.

Publications

Gecko was presented in Interspeech 2019, the world's leading Speech Technology conference. See this video for an overview and the accepted paper.

Citation

If you use Gecko please use the following

    @inproceedings{Gecko2019,
      Author = {Golan Levy, Raquel Sitman, Ido Amir, Eduard Golshtein, Ran Mochary, Eilon Reshef, Reichart, Omri Allouche},
      Title = {GECKO - A Tool for Effective Annotation of Human Conversations},
      Booktitle = {20th Annual Conference of the International Speech Communication Association, Interspeech 2019},
      Year = {2019},
      Month = {September},
      Address = {Herzliya, Israel},
      Url = {https://github.com/gong-io/gecko/blob/master/docs/gecko_interspeech_2019_paper.pdf}
    }

GitHub