Skip to main content

Evaluate the quality of SRT files using the multilingual multimodal SONAR model.

Project description

SONAR Subtitling Evaluator

Code to evaluate the quality of SRT files using the multilingual multimodal SONAR sentence embedding model.

The evaluation accounts for the semantic similarity (computed as a cosine similarity) between each subtitle block and the corresponding audio to which the block is assigned to (through the timestamps in the SRT). The returned scores range in [-1, 1] where the higher, the better.


Ensure that you have libsndfile installed in you environment. Then, run:

pip install SubSONAR

or, in the source root of this repository:

pip install -e .

The installation has been tested with python 3.8 and 3.10.


Example usage for Italian SRTs and English audios of two files (1 and 2):

subsonar \
  --srt-files \
  --audio-files 1.wav 2.wav \
  --text-lang ita_Latn --audio-lang eng \
  -bs 32

Please set the batch size bs according to your GPU capacity.

The available languages for the speech encoder (--audio-lang) can be found in the SONAR repository, while the text encoder (--text-lang) supports the 200 languages of NLLB.


SONAR Subtitling Evaluator is licensed under Apache Version 2.0.

However, the SONAR encoders have a dedicated license that can be found in their repository LICENSE. Please check the license for the encoders you are using.


If you find this project useful, please cite:

title = {{SBAAM! Eliminating Transcript Dependency in Automatic Subtitling}},
author = {Gaido, Marco and Papi, Sara and Negri, Matteo and Cettolo, Mauro and Bentivogli, Luisa},
booktitle = "Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
year = "2024",
address = "Bangkok, Thailand",

Project details

Release history Release notifications | RSS feed

This version


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

subsonar-1.0.tar.gz (11.7 kB view hashes)

Uploaded Source

Built Distribution

SubSONAR-1.0-py3-none-any.whl (12.5 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page