The MediaPipe Language Detector task lets you identify the language of a piece of text. This task operates on text data with a machine learning (ML) model and outputs a list of predictions, where each prediction consists of an [ISO 639-1](https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes) language code and a probability. [Open in Web Demos](https://google-ai-edge.github.io/mediapipe-samples-web/#/text/language_detector)[Edit in StackBlitz](https://stackblitz.com/github/google-ai-edge/mediapipe-samples-web?file=src/tasks/language-detector.ts)
## Quickstart Install the SDK for your platform, initialize the task with a model, and run inference. ### Web npm`npm install @mediapipe/tasks-text`[Web demo](https://google-ai-edge.github.io/mediapipe-samples-web/#/text/language_detector)·[Edit in StackBlitz](https://stackblitz.com/github/google-ai-edge/mediapipe-samples-web?file=src/tasks/language-detector.ts)·[Setup guide](https://developers.google.com/edge/mediapipe/solutions/setup_web.md.txt) ```typescript import { FilesetResolver, LanguageDetector } from "@mediapipe/tasks-text"; const MODEL_URL = "https://storage.googleapis.com/mediapipe-models/language_detector/language_detector/float32/1/language_detector.tflite"; const textFiles = await FilesetResolver.forTextTasks( "https://cdn.jsdelivr.net/npm/@mediapipe/tasks-text@latest/wasm" ); const languageDetector = await LanguageDetector.createFromOptions(textFiles, { baseOptions: { modelAssetPath: MODEL_URL }, }); const inputText = "Bonjour tout le monde, comment allez-vous ?"; const detectionResult = languageDetector.detect(inputText); ``` ### Android (Kotlin) Gradle`implementation("com.google.mediapipe:tasks-text:latest.release")`[Sample code](https://github.com/google-ai-edge/mediapipe-samples/tree/main/examples/language_detector/android)·[Setup guide](https://developers.google.com/edge/mediapipe/solutions/setup_android.md.txt) ```kotlin import com.google.mediapipe.tasks.core.BaseOptions import com.google.mediapipe.tasks.text.languagedetector.LanguageDetector import com.google.mediapipe.tasks.text.languagedetector.LanguageDetector.LanguageDetectorOptions val baseOptions = BaseOptions.builder() .setModelAssetPath("language_detector.tflite") .build() val options = LanguageDetectorOptions.builder() .setBaseOptions(baseOptions) .setScoreThreshold(0.5f) .build() val languageDetector = LanguageDetector.createFromOptions(context, options) val inputText = "Bonjour tout le monde, comment allez-vous ?" val result = languageDetector.detect(inputText) ``` ### Android (Java) Gradle`implementation 'com.google.mediapipe:tasks-text:latest.release'`[Sample code](https://github.com/google-ai-edge/mediapipe-samples/tree/main/examples/language_detector/android)·[Setup guide](https://developers.google.com/edge/mediapipe/solutions/setup_android.md.txt) ```java import com.google.mediapipe.tasks.core.BaseOptions; import com.google.mediapipe.tasks.text.languagedetector.LanguageDetector; import com.google.mediapipe.tasks.text.languagedetector.LanguageDetector.LanguageDetectorOptions; import com.google.mediapipe.tasks.text.languagedetector.LanguageDetectorResult; BaseOptions baseOptions = BaseOptions.builder() .setModelAssetPath("language_detector.tflite") .build(); LanguageDetectorOptions options = LanguageDetectorOptions.builder() .setBaseOptions(baseOptions) .setScoreThreshold(0.5f) .build(); LanguageDetector languageDetector = LanguageDetector.createFromOptions(context, options); String inputText = "Bonjour tout le monde, comment allez-vous ?"; LanguageDetectorResult result = languageDetector.detect(inputText); ``` ### iOS SwiftPM`https://github.com/google-ai-edge/mediapipe → MediaPipeTasksText`[Package.swift](https://developers.google.com/edge/mediapipe/solutions/setup_ios.md.txt#package_swift)·[Sample code](https://github.com/google-ai-edge/mediapipe-samples/tree/main/examples/language_detector/ios)·[Setup guide](https://developers.google.com/edge/mediapipe/solutions/setup_ios.md.txt) ```swift import Foundation import MediaPipeTasksText guard let modelPath = Bundle.main.path(forResource: "language_detector", ofType: "tflite") else { return } let options = LanguageDetectorOptions() options.baseOptions.modelAssetPath = modelPath options.scoreThreshold = 0.5 let languageDetector = try LanguageDetector(options: options) let inputText = "Bonjour tout le monde, comment allez-vous ?" let result = try languageDetector.detect(text: inputText) ``` ### Python pip`pip install mediapipe`[Sample code](https://colab.research.google.com/github/google-ai-edge/mediapipe-samples/blob/main/examples/language_detector/python/%5BMediaPipe_Python_Tasks%5D_Language_Detector.ipynb)·[Setup guide](https://developers.google.com/edge/mediapipe/solutions/setup_python.md.txt) ```python import mediapipe as mp BaseOptions = mp.tasks.BaseOptions LanguageDetector = mp.tasks.text.LanguageDetector LanguageDetectorOptions = mp.tasks.text.LanguageDetectorOptions model_path = "language_detector.tflite" options = LanguageDetectorOptions(base_options=BaseOptions(model_asset_path=model_path), score_threshold=0.5) input_text = "Bonjour tout le monde, comment allez-vous ?" with LanguageDetector.create_from_options(options) as detector: detection_result = detector.detect(input_text) ``` ### Flutter pub`flutter pub add mediapipe_text`[Live gallery](https://google.github.io/flutter-mediapipe/)·[Sample code](https://github.com/google/flutter-mediapipe/tree/main/samples/text/language_detector)·[Setup guide](https://developers.google.com/edge/mediapipe/solutions/setup_flutter.md.txt) ```dart import 'package:mediapipe_text/mediapipe_text.dart'; Future detectLanguage(String text) async { final detector = await LanguageDetector.create( LanguageDetectorOptions(model: TextModels.languageDetector), ); try { final result = await detector.detect(text); for (final prediction in result.predictions.take(3)) { print('${prediction.languageCode}: ${prediction.probability}'); } } finally { await detector.dispose(); } } ``` ## Task details This section describes the capabilities, inputs, outputs, and configuration options of this task. ### Features - **Score threshold** - Filter results based on prediction scores - **Label allowlist and denylist** - Specify the categories detected Inputs Language Detector accepts the following input data type: - String Outputs The Language Detector outputs a list of predictions containing: - Language code: An [ISO 639-1](https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes) language / locale code (e.g. "en" for English, "uz" for Uzbek, "ja-Latn" for Japanese (romaji)) as a string. - Probability: the confidence score for this prediction, expressed as a probability between zero and one as floating point value. ### Configuration options This task has the following configuration options: ### Web | Option Name | Description | Value Range | Default Value | |---|---|---|---| | `displayNamesLocale` | Locale code for display names provided in the model metadata, if available. | Locale code | `"en"` | | `maxResults` | Maximum number of top-scored language predictions to return (`-1` returns all results). | Integer | `-1` | | `scoreThreshold` | Prediction score threshold overriding the model metadata default. Predictions lower than this score are rejected. | Float `[0.0, 1.0]` | Not set | | `categoryAllowlist` | Optional list of allowed BCP 47 language codes. Mutually exclusive with `categoryDenylist`. | List of strings | Not set | | `categoryDenylist` | Optional list of excluded BCP 47 language codes. Mutually exclusive with `categoryAllowlist`. | List of strings | Not set | ### Android | Option Name | Description | Value Range | Default Value | |---|---|---|---| | `displayNamesLocale` | Locale code for display names provided in the model metadata, if available. | Locale code | `"en"` | | `maxResults` | Maximum number of top-scored language predictions to return (`-1` returns all results). | Integer | `-1` | | `scoreThreshold` | Prediction score threshold overriding the model metadata default. Predictions lower than this score are rejected. | Float `[0.0, 1.0]` | Not set | | `categoryAllowlist` | Optional list of allowed BCP 47 language codes. Mutually exclusive with `categoryDenylist`. | List of strings | Not set | | `categoryDenylist` | Optional list of excluded BCP 47 language codes. Mutually exclusive with `categoryAllowlist`. | List of strings | Not set | ### iOS | Option Name | Description | Value Range | Default Value | |---|---|---|---| | `displayNamesLocale` | Locale code for display names provided in the model metadata, if available. | Locale code | `"en"` | | `maxResults` | Maximum number of top-scored language predictions to return (`-1` returns all results). | Integer | `-1` | | `scoreThreshold` | Prediction score threshold overriding the model metadata default. Predictions lower than this score are rejected. | Float `[0.0, 1.0]` | Not set | | `categoryAllowlist` | Optional list of allowed BCP 47 language codes. Mutually exclusive with `categoryDenylist`. | List of strings | Not set | | `categoryDenylist` | Optional list of excluded BCP 47 language codes. Mutually exclusive with `categoryAllowlist`. | List of strings | Not set | ### Python | Option Name | Description | Value Range | Default Value | |---|---|---|---| | `display_names_locale` | Locale code for display names provided in the model metadata, if available. | Locale code | `"en"` | | `max_results` | Maximum number of top-scored language predictions to return (`-1` returns all results). | Integer | `-1` | | `score_threshold` | Prediction score threshold overriding the model metadata default. Predictions lower than this score are rejected. | Float `[0.0, 1.0]` | Not set | | `category_allowlist` | Optional list of allowed BCP 47 language codes. Mutually exclusive with `category_denylist`. | List of strings | Not set | | `category_denylist` | Optional list of excluded BCP 47 language codes. Mutually exclusive with `category_allowlist`. | List of strings | Not set | ### Flutter | Option Name | Description | Value Range | Default Value | |---|---|---|---| | `displayNamesLocale` | Locale code for display names provided in the model metadata, if available. `null` keeps Google's default (`"en"`). | Locale code | `null` | | `maxResults` | Maximum number of top-scored language predictions to return (any negative value returns all results; `0` is rejected). | Integer other than `0` | `-1` | | `scoreThreshold` | Prediction score threshold. Predictions lower than this score are rejected. The value is always sent, so the default `0` keeps every language rather than applying the model metadata's threshold. | Float `[0.0, 1.0]` | `0` | | `categoryAllowlist` | Optional list of allowed BCP 47 language codes. Mutually exclusive with `categoryDenylist`. | `List` | Empty | | `categoryDenylist` | Optional list of excluded BCP 47 language codes. Mutually exclusive with `categoryAllowlist`. | `List` | Empty | ### Task output Each `LanguageDetectorResult` returns a list of `languages` (`LanguagePrediction`) ordered by descending probability (in Dart, the list is named `predictions`): | Field | Type / Range | Description | |---|---|---| | `languageCode` | `string` (BCP-47 / ISO 639-1) | Predicted language code, such as `"fr"`, `"en"`, `"ja"`, or `"und"` when below threshold (`language_code` in Python). | | `probability` | `float` (`[0.0, 1.0]`) | Confidence score for the predicted language. | #### Example `LanguageDetectorResult` payload The following example shows the language prediction for `"Il y a beaucoup de bouches qui parlent et fort peu de têtes qui pensent."`: ```javascript { languages: [ { languageCode: "fr", probability: 0.9998 } ] } ``` ## Models ![Language Detector pipeline overview](https://developers.google.com/static/edge/mediapipe/images/solutions/diagrams/language_detector.svg) We offer a default, recommended model when you start developing with this task. ### Language detector model (recommended) This model is built to be lightweight (315 KB) and uses embedding-based, neural network classification architecture. The model identifies language using an [ISO 639-1](https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes) language code, and can identify 110 languages. For a list of languages supported by the model, see the [label file](https://storage.googleapis.com/mediapipe-tasks/language_detector/labels.txt), which lists languages by their ISO 639-1 code. | Model name | Input shape | Quantization type | Model card | Versions | |---|---|---|---|---| | [Language Detector](https://storage.googleapis.com/mediapipe-models/language_detector/language_detector/float32/1/language_detector.tflite) | string UTF-8 | none (float32) | [info](https://storage.googleapis.com/mediapipe-assets/LanguageDetector%20Model%20Card.pdf) | [Latest](https://storage.googleapis.com/mediapipe-models/language_detector/language_detector/float32/1/language_detector.tflite) | ## Task benchmarks Here are the task benchmarks for the whole pipeline based on the preceding pre-trained models. The latency result is the average latency on Pixel 6 using CPU / GPU. | Model name | Device | CPU latency | GPU latency | |---|---|---|---| | Language Detector | Pixel 6 | 0.31ms | - |