Lingua Language Detector

2.2.0 · active · verified Sat Apr 11

Lingua Language Detector is an accurate natural language detection library for Python, suitable for both short text snippets and mixed-language texts. It leverages Rust bindings for high performance and low memory consumption, supporting 75 languages offline. The current version is 2.2.0, with an active development cycle featuring regular minor and patch releases.

Warnings

breaking Version 2.0.0 completely replaced the pure Python implementation with Rust bindings, changing the underlying API. Code written for `1.x` versions will likely break and require adaptation.
Fix: Review the official documentation for `lingua-language-detector` 2.x and update code to use the new API, especially `LanguageDetectorBuilder` and `Language` enum.
breaking Starting with version 2.1.1, support for Python 3.10 and 3.11 was dropped. The library now requires Python 3.12 or newer.
Fix: Upgrade your Python environment to version 3.12 or higher.
gotcha Using `LanguageDetectorBuilder.with_low_accuracy_mode()` improves performance and reduces memory but significantly decreases detection accuracy for texts shorter than 120 characters.
Fix: Avoid `with_low_accuracy_mode()` for short texts or applications where high accuracy is paramount. Consider pre-filtering languages to optimize performance without sacrificing accuracy.
gotcha The `detect_multiple_languages_of()` method for mixed-language texts is considered experimental. Its results are highly dependent on the input text and it performs best in high-accuracy mode with longer words.
Fix: Be aware of potential inaccuracies when using `detect_multiple_languages_of()` on very short or ambiguous mixed texts. Test thoroughly with your specific data.

Install

pip install lingua-language-detector Install stable version

Imports

Language
```
from lingua import Language
```

LanguageDetectorBuilder

from lingua import LanguageDetectorBuilder

Quickstart

This quickstart demonstrates how to build a language detector, detect the language of a single text, and identify multiple languages within a mixed-language text. It utilizes the `Language` enum and `LanguageDetectorBuilder` to configure and create a detector instance.

from lingua import Language, LanguageDetectorBuilder

# Build a detector for specific languages
languages = [Language.ENGLISH, Language.FRENCH, Language.GERMAN]
detector = LanguageDetectorBuilder.from_languages(*languages).build()

# Detect a single language
text_single = "languages are awesome"
detected_language_single = detector.detect_language_of(text_single)
print(f"Detected language (single): {detected_language_single.name}")

# Detect multiple languages in mixed text (experimental)
text_mixed = "Hello world, comment ça va? Das ist ein Test."
detected_languages_mixed = detector.detect_multiple_languages_of(text_mixed)
print("Detected languages (mixed):")
for result in detected_languages_mixed:
    print(f"  - {result.language.name}: '{text_mixed[result.start_index:result.end_index]}' ({result.start_index}-{result.end_index})")

view raw JSON →