Back to main demo page

asr demo

Recognise Finnish speech with Kaldi and Aalto-asr
Show help


Entering input

Currently, only file uploads are supported. Any format known to ffmpeg may work, but wav and mp3 have been tested.

Understanding output

The audio is split into chunks separated by silence. These chunks are processed separately, in parallel. The output shows them in the correct order. Tabular output shows

  1. The full recognized text, once it is ready
  2. The recognized chunks, as they are completed
  3. A table with each word in the chunk, with time information

When results are complete, a tsv file with all the timing information is generated for downloading.

Page generated in 0.01 seconds