whatisgithub

What is sonar?

facebookresearch/sonar — explained in plain English

Analysis updated 2026-07-17 · repo last pushed 2025-10-10

★ 896PythonAudience · researcherComplexity · 4/5QuietSetup · moderate

In one sentence

SONAR converts sentences from text or speech, across 200 languages, into a shared numerical space so you can compare, translate, or measure similarity between them regardless of language or input type.

Mindmap

mindmap
  root((repo))
    What it does
      Encodes text and speech
      Shared embedding space
      Cross-language comparison
    Tech stack
      Python
      Text encoders
      Speech encoders
    Use cases
      Multilingual translation
      Cross-language search
      Toxicity detection
    Audience
      ML researchers
      Translation engineers

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

What do people build with it?

USE CASE 1

Build a multilingual chatbot that understands support tickets in 200 languages without separate models per language.

USE CASE 2

Score machine translation quality automatically using the included BLASER 2.0 evaluation models.

USE CASE 3

Search or match similar sentences across different languages using a shared embedding space.

USE CASE 4

Detect toxicity in multilingual speech using SONAR's speech encoders.

What is it built with?

PythonPyTorch

How does it compare?

facebookresearch/sonarywnd1144/gopay_plus_automaticdjango/django-localflavor
Stars896918919
LanguagePythonPythonPython
Last pushed2025-10-10—2026-06-23
MaintenanceQuiet—Active
Setup difficultymoderatehardeasy
Complexity4/54/52/5
Audienceresearchergeneraldeveloper

Figures from each repo's GitHub metadata at analysis time.

How do you get it running?

Difficulty · moderate Time to first run · 1h+

Requires downloading pretrained encoder/decoder models and a Python ML environment with PyTorch.

Copy-paste prompts

Prompt 1
Show me how to encode a sentence with SONAR and compare it to a sentence in another language.
Prompt 2
Help me use SONAR embeddings to evaluate machine translation quality with BLASER 2.0.
Prompt 3
Explain how SONAR's text and speech encoders share the same embedding space.
Prompt 4
How do I set up SONAR to power cross-language search across 200 supported languages?

Frequently asked questions

What is sonar?

SONAR converts sentences from text or speech, across 200 languages, into a shared numerical space so you can compare, translate, or measure similarity between them regardless of language or input type.

What language is sonar written in?

Mainly Python. The stack also includes Python, PyTorch.

Is sonar actively maintained?

Quiet — no commits in 6-12 months (last push 2025-10-10).

How hard is sonar to set up?

Setup difficulty is rated moderate, with roughly 1h+ to a first successful run.

Who is sonar for?

Mainly researcher.

Open on GitHub → Ask about another repo

This repo across BitVibe Labs

Verify against the repo before relying on details.

What are similar repos?