audio-transcriber-docker

日本語のREADMEはこちらから

Automatically transcribe the audio of video / audio files using Speech Recognition.

Warning / Disclaimer

The author is not responsible for any problems caused by the user using this project.

Or rather, if you want to use it seriously, use Google Cloud Platform Speech-to-Text.
We do not provide any support.

Installation

1. Git Clone

git clone https://github.com/book000/audio-transcriber-docker.git

2. Run start-mount.bat or start-mount.sh (First)

Run start-mount.bat on Windows, start-mount.sh on Linux.
These programs mount the user-dir folder and output folder.

If you pay attention only to DISPLAY, you can just enter the other questions.

3. Permit using microphone

As an initial setting, you need to set the microphone permission and the microphone to be used. If the audio is not recognized correctly the first time, you need to perform this work again.

3-1. Press Allow to allow

3-2. Press the video symbol and change the Microphone field to Virtual Source Virtual Mic on Monitor of Virtual_Microphone_Output

4. Close the window

Press Ctrl+C at the terminal to abort the process. Run docker stop audio-transcriber, docker rm audio-transcriber if necessary.

Usage

After Installation, if you use it normally

1. Run start.bat or start.sh

Run start.bat on Windows, start.sh on Linux.
These programs mount the output folder.

The questions that come up are below.

• Movie or Audio File Path
  • Specify the files to be loaded (video and audio files). mp4, m4v, webm and anything else that can be converted with ffmpeg
• DockerName
  • Specify the name of the container in Docker. Note that if you specify an existing container name, it will be stop and rm without checking.
  • Use audio-transcriber if you enter without specifying.
• DISPLAY
  • Specifies the DISPLAY of the host (e.g., 192.168.0.100:0) because there is no video output destination in the Docker.

Make a destination for X11 with TeraTerm or Mobaxterm.

2. Processing

After Chrome is launched, it works.

3. View the output/{DateTime}.txt

It is written to output/{DateTime}.txt.

License

The license for this project is MIT License.