Sign inSign up

gianpaolocoro/whisperx_aligner

By gianpaolocoro

•Updated almost 2 years ago

Speech aligner based on whisperX for experimentations.

Image
0

279

gianpaolocoro/whisperx_aligner repository overview

⁠A WhisperX-based laboratory for Automatic Speech Recognition and Alignment

Sample file: https://data.d4science.net/53BM⁠

Optimal input audio format: WAV mono, 16 bit, 16kHz

Sample run (on audio.wav input, executed from the folder containing the wave file):

docker run --runtime=nvidia --gpus all \
 -v ${PWD}:/home/docker/ \
 -it gianpaolocoro/whisperx_aligner \
 bash -c "source /root/anaconda3/etc/profile.d/conda.sh && \
 conda activate whisperx && cd /whisperX/ && \
 export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/root/anaconda3/pkgs/pytorch-2.0.0-py3.10_cuda11.8_cudnn8.7.0_0/lib/python3.10/site-packages/torch/lib/ && \
 whisperx /home/docker/audio.wav --compute_type int8 --language it && \
 cp /whisperX/audio.* /home/docker/ && \
 chmod 777 /home/docker/audio.*"

Higher accuracy model using intensive GPU computation:

docker run --runtime=nvidia --gpus all \
-v ${PWD}:/home/docker/ \
-it gianpaolocoro/whisperx_aligner \
bash -c "source /root/anaconda3/etc/profile.d/conda.sh && \
conda activate whisperx && cd /whisperX/ && \
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/root/anaconda3/pkgs/pytorch-2.0.0-py3.10_cuda11.8_cudnn8.7.0_0/lib/python3.10/site-packages/torch/lib/ && \
whisperx /home/docker/audio.wav --compute_type int8 --language it --model large-v3 && \
cp /whisperX/audio.* /home/docker/ && \
chmod 777 /home/docker/audio.*"

Tag summary

Content type

Image

Digest

sha256:82cf135fd…

Size

25.4 GB

Last updated

almost 2 years ago

docker pull gianpaolocoro/whisperx_aligner