https://github.com/mcdallas/whispersub

format whisper transcripts to .srt
https://github.com/mcdallas/whispersub

openai srt subtitles transcription whisper whisper-cpp

Last synced: 3 months ago
JSON representation

format whisper transcripts to .srt

• Host: GitHub
• URL: https://github.com/mcdallas/whispersub
• Owner: mcdallas
• Created: 2023-07-15T02:17:10.000Z (over 1 year ago)
• Default Branch: master
• Last Pushed: 2023-07-15T02:26:37.000Z (over 1 year ago)
• Last Synced: 2024-10-15T03:06:31.519Z (3 months ago)
• Topics: openai, srt, subtitles, transcription, whisper, whisper-cpp
• Language: Rust
• Homepage:
• Size: 3.91 KB
• Stars: 15
• Watchers: 2
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        

## whispersub

A dead simple utility to format the output of OpenAI's whisper model (or whisper.cpp) into an .srt file.

### Usage

``` bash

whispersub input.txt -o output.srt

```

you can also pipe the output of whisper.cpp into whispersub

```bash

whisper-cpp --file audio.wav --language en --model ggml-medium.en.bin | whispersub

```

or use a little hellper function to extract the audio from a video, pipe it to whisper.cpp and then to whispersub

```bash

makesub () {

    filename=$(basename -- "$1")

    filename="${filename%.*}"

    model=${HOME}/.local/share/whisper/ggml-medium.en.bin

    ffmpeg -i "$1" -vn -acodec pcm_s16le -ar 16000 -ac 2 -f wav - | 

    nice -n 20 whisper-cpp --threads "$(nproc)" --file - --language en --model "$model" |

    whispersub -o "${filename}.en.srt"

}

makesub video.mp4

```