transcriber #11
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| name: "transcriber" | |
| on: | |
| workflow_dispatch: | |
| inputs: | |
| audioURL: | |
| description: "Audio file URL" | |
| required: true | |
| type: string | |
| whisperModel: | |
| description: "The model size for whisper, default is large-v2. For more details, checkout https://github.com/openai/whisper#available-models-and-languages" | |
| default: "large-v2" | |
| type: string | |
| targetLanguage: | |
| description: "The target language for translation, default is zh (Chinese)" | |
| default: "zh" | |
| type: string | |
| translate: | |
| description: "Whether translating the srt file" | |
| default: true | |
| type: boolean | |
| jobs: | |
| transcribe: | |
| name: "Transcribe" | |
| runs-on: ubuntu-latest | |
| container: ghcr.io/${{ github.repository }}:${{ github.ref_name }} | |
| env: | |
| audioURL: ${{ inputs.audioURL }} | |
| steps: | |
| - name: Checkout repo | |
| uses: actions/checkout@v3 | |
| - name: Install packages | |
| run: apt update && apt install -y ffmpeg fonts-noto-cjk curl file | |
| - name: Download / convert audio file | |
| run: | | |
| mkdir -p workdir | |
| cd workdir | |
| echo "Downloading file..." | |
| filename=$(curl -sJOL ${audioURL} -w "%{filename_effective}") | |
| filename_base=${filename%.*} | |
| echo "filename=${filename}" >> $GITHUB_ENV | |
| echo "filename_base=${filename_base}" >> $GITHUB_ENV | |
| if [[ $(file -bi "$filename") == video/* ]]; then # Determine filetype | |
| echo "Detected input as a video file, converting to .wav..." | |
| ffmpeg -i "$filename" "$filename_base".wav | |
| is_video=true | |
| whisper_file="${filename_base}.wav" | |
| else | |
| is_video=false | |
| whisper_file="${filename}" | |
| fi | |
| echo "is_video=${is_video}" >> $GITHUB_ENV | |
| echo "whisper_file=${whisper_file}" >> $GITHUB_ENV | |
| echo "Downloaded file: ${filename}" | |
| - name: Transcribe! | |
| working-directory: ./workdir | |
| run: | | |
| export PATH="/usr/src/.venv/bin:$PATH" | |
| if [[ ${{ inputs.translate }} == true ]]; then | |
| /usr/src/.venv/bin/python /usr/src/src/main.py -d cpu -t -m ${{ inputs.whisperModel }} -tl ${{ inputs.targetLanguage }} ${{ env.whisper_file }} | |
| else | |
| /usr/src/.venv/bin/python /usr/src/src/main.py -d cpu -m ${{ inputs.whisperModel }} -tl ${{ inputs.targetLanguage }} ${{ env.whisper_file }} | |
| fi | |
| - name: Generate videos with/without subtitle | |
| working-directory: ./workdir | |
| run: | | |
| if [[ ${{ env.is_video }} == true ]]; then # Determine filetype | |
| echo "Detected input as a video file, skipping video generation..." | |
| echo "Embedding subtitles..." | |
| ffmpeg -i ${{ env.filename }} -vf subtitles=${{ env.filename_base }}_${{ inputs.targetLanguage }}.srt ${{ env.filename_base }}_${{ inputs.targetLanguage }}.mp4 | |
| else | |
| echo "Generating videos using ffmpeg..." | |
| ffmpeg -f lavfi -i color=c=black:s=1920x1080:r=5 -i ${{ env.filename }} -crf 0 -c:a copy -shortest ${{ env.filename_base }}.orig.mp4 | |
| echo "Embedding subtitles..." | |
| ffmpeg -i ${{ env.filename_base }}.orig.mp4 -vf subtitles=${{ env.filename_base }}_${{ inputs.targetLanguage }}.srt ${{ env.filename_base }}_${{ inputs.targetLanguage }}.mp4 | |
| echo "Deleting redundant files..." | |
| rm -v ${{ env.filename_base }}.orig.mp4 | |
| fi | |
| - name: Upload artifact | |
| uses: actions/upload-artifact@v3 | |
| with: | |
| path: workdir/ | |
| name: ${{ env.filename_base }} |