Skip to content

FunASR

Offline audio and video transcription with punctuated text

Visit official site

What FunASR is

FunASR is described as an offline file transcription service package that integrates voice activity detection, speech recognition and punctuation models. The package turns long audio and video into punctuated text and handles multiple transcription requests at the same time.

What you can do with it

  • Run a full recognition chain with voice activity detection, speech recognition and punctuation prediction
  • Transcribe dozens of hours of long audio and video into punctuated text
  • Handle hundreds of concurrent transcription requests
  • Use character-level timestamps to align text with speech

Application scenarios

  • An enterprise transcribes meeting recordings to produce meeting notes
  • An online education platform turns lecture audio into text material for review
  • A media company converts interview recordings into text

Getting started

  • Install docker if it is not already present
  • Pull the docker image of the FunASR package
  • Start the docker image and map the resource directories
  • Launch the funasr-wss-server service inside docker

Pros & cons

✓ What we like

  • ffmpeg is integrated on the server side for many audio and video formats
  • Character-level timestamps help align text with speech
  • ITN and custom hotwords are supported

! What to watch out for

  • Setup requires docker and mapped resource directories
  • The service is delivered as an offline file transcription package
  • Recognition depends on the bundled detection, recognition and punctuation models

FAQ

What does the service output?

The material states long audio and video are converted into punctuated text.

How is it deployed?

The steps name installing docker, pulling the image, mapping directories and starting the funasr-wss-server service.

Last reviewed: 2026-11-15

More AI audio tools tools

View all →

How we review