Skip to content

Puhti and Mahti computing services have been decommissioned. Puhti and Mahti login nodes and storage services will remain available until 15 October 2026, but are no longer covered by service contracts. Please clean up and migrate your data to Roihu ASAP. See Roihu data migration guide for instructions.

Whisper

Whisper is a general-purpose speech recognition model. It is trained on a large dataset of diverse audio and is also a multitasking model that can perform multilingual speech recognition, speech translation, and language identification.

Available

Faster-Whisper-XXL r245.4 is available in Puhti.

License

Faster-Whisper-XXL is licenced using MIT licence.

Usage

CSC users can easily install Whisper in their own Python virtual environments in Puhti and Mahti. In addition, Puhti has a pre-installed Faster-Whisper-XXL version of Whisper. This Whisper environment can be activated in Puhti with command:

module load whisper

Sample command:

whisper audio.mp3 --model medium

Sample command with diarization enabled:

whisper interview.mp4 --model large --language French --threads 4 --diarize pyannote_v3.0 --diarize_threads 4 --num_speakers 2 -o interview_results

Example batch script

Whisper can use utilize GPU computing effectively. The example batch script below reserves one GPU for a Whisper job.

#!/bin/bash
#SBATCH --account=<project>
#SBATCH --partition=gpu
#SBATCH --nodes=1
#SBATCH --ntasks=1
#SBATCH --cpus-per-task=1
#SBATCH --mem=8G
#SBATCH --time=1:00:00
#SBATCH --gres=gpu:v100:1

module load whisper
srun whisper interview.mp4 --model large --language French --threads 4 --diarize pyannote_v3.0 --diarize_threads 4 --num_speakers 2 -o interview_results

More information