Specify source language for speech-to-text

  • Article

In this article, you'll learn how to specify the source language for an audio input passed to the Speech SDK for speech recognition. The example code that's provided specifies a custom speech model for improved recognition.

Specify source language in C#

In the following example, the source language is provided explicitly as a parameter by using the SpeechRecognizer construct:

var recognizer = new SpeechRecognizer(speechConfig, "de-DE", audioConfig);

In the following example, the source language is provided by using SourceLanguageConfig. Then, sourceLanguageConfig is passed as a parameter to the SpeechRecognizer construct.

var sourceLanguageConfig = SourceLanguageConfig.FromLanguage("de-DE");
var recognizer = new SpeechRecognizer(speechConfig, sourceLanguageConfig, audioConfig);

In the following example, the source language and custom endpoint are provided by using SourceLanguageConfig. Then, sourceLanguageConfig is passed as a parameter to the SpeechRecognizer construct.

var sourceLanguageConfig = SourceLanguageConfig.FromLanguage("de-DE", "The Endpoint ID for your custom model.");
var recognizer = new SpeechRecognizer(speechConfig, sourceLanguageConfig, audioConfig);

Note

The SpeechRecognitionLanguage and EndpointId set methods are deprecated from the SpeechConfig class in C#. The use of these methods is discouraged. Don't use them when you create a SpeechRecognizer construct.

Specify source language in C++

In the following example, the source language is provided explicitly as a parameter by using the FromConfig method.

auto recognizer = SpeechRecognizer::FromConfig(speechConfig, "de-DE", audioConfig);

In the following example, the source language is provided by using SourceLanguageConfig. Then, sourceLanguageConfig is passed as a parameter to FromConfig when you create the recognizer construct.

auto sourceLanguageConfig = SourceLanguageConfig::FromLanguage("de-DE");
auto recognizer = SpeechRecognizer::FromConfig(speechConfig, sourceLanguageConfig, audioConfig);

In the following example, the source language and custom endpoint are provided by using SourceLanguageConfig. Then, sourceLanguageConfig is passed as a parameter to FromConfig when you create the recognizer construct.

auto sourceLanguageConfig = SourceLanguageConfig::FromLanguage("de-DE", "The Endpoint ID for your custom model.");
auto recognizer = SpeechRecognizer::FromConfig(speechConfig, sourceLanguageConfig, audioConfig);

Note

SetSpeechRecognitionLanguage and SetEndpointId are deprecated methods from the SpeechConfig class in C++ and Java. The use of these methods is discouraged. Don't use them when you create a SpeechRecognizer construct.

Specify source language in Java

In the following example, the source language is provided explicitly when you create a new SpeechRecognizer construct.

SpeechRecognizer recognizer = new SpeechRecognizer(speechConfig, "de-DE", audioConfig);

In the following example, the source language is provided by using SourceLanguageConfig. Then, sourceLanguageConfig is passed as a parameter when you create a new SpeechRecognizer construct.

SourceLanguageConfig sourceLanguageConfig = SourceLanguageConfig.fromLanguage("de-DE");
SpeechRecognizer recognizer = new SpeechRecognizer(speechConfig, sourceLanguageConfig, audioConfig);

In the following example, the source language and custom endpoint are provided by using SourceLanguageConfig. Then, sourceLanguageConfig is passed as a parameter when you create a new SpeechRecognizer construct.

SourceLanguageConfig sourceLanguageConfig = SourceLanguageConfig.fromLanguage("de-DE", "The Endpoint ID for your custom model.");
SpeechRecognizer recognizer = new SpeechRecognizer(speechConfig, sourceLanguageConfig, audioConfig);

Note

setSpeechRecognitionLanguage and setEndpointId are deprecated methods from the SpeechConfig class in C++ and Java. The use of these methods is discouraged. Don't use them when you create a SpeechRecognizer construct.

Specify source language in Python

In the following example, the source language is provided explicitly as a parameter by using the SpeechRecognizer construct.

speech_recognizer = speechsdk.SpeechRecognizer(
        speech_config=speech_config, language="de-DE", audio_config=audio_config)

In the following example, the source language is provided by using SourceLanguageConfig. Then, SourceLanguageConfig is passed as a parameter to the SpeechRecognizer construct.

source_language_config = speechsdk.languageconfig.SourceLanguageConfig("de-DE")
speech_recognizer = speechsdk.SpeechRecognizer(
        speech_config=speech_config, source_language_config=source_language_config, audio_config=audio_config)

In the following example, the source language and custom endpoint are provided by using SourceLanguageConfig. Then, SourceLanguageConfig is passed as a parameter to the SpeechRecognizer construct.

source_language_config = speechsdk.languageconfig.SourceLanguageConfig("de-DE", "The Endpoint ID for your custom model.")
speech_recognizer = speechsdk.SpeechRecognizer(
        speech_config=speech_config, source_language_config=source_language_config, audio_config=audio_config)

Note

The speech_recognition_language and endpoint_id properties are deprecated from the SpeechConfig class in Python. The use of these properties is discouraged. Don't use them when you create a SpeechRecognizer construct.

Specify source language in JavaScript

The first step is to create a SpeechConfig construct:

var speechConfig = sdk.SpeechConfig.fromSubscription("YourSubscriptionkey", "YourRegion");

Next, specify the source language of your audio with speechRecognitionLanguage:

speechConfig.speechRecognitionLanguage = "de-DE";

If you're using a custom model for recognition, you can specify the endpoint with endpointId:

speechConfig.endpointId = "The Endpoint ID for your custom model.";

Specify source language in Objective-C

In the following example, the source language is provided explicitly as a parameter by using the SPXSpeechRecognizer construct.

SPXSpeechRecognizer* speechRecognizer = \
    [[SPXSpeechRecognizer alloc] initWithSpeechConfiguration:speechConfig language:@"de-DE" audioConfiguration:audioConfig];

In the following example, the source language is provided by using SPXSourceLanguageConfiguration. Then, SPXSourceLanguageConfiguration is passed as a parameter to the SPXSpeechRecognizer construct.

SPXSourceLanguageConfiguration* sourceLanguageConfig = [[SPXSourceLanguageConfiguration alloc]init:@"de-DE"];
SPXSpeechRecognizer* speechRecognizer = [[SPXSpeechRecognizer alloc] initWithSpeechConfiguration:speechConfig
                                                                     sourceLanguageConfiguration:sourceLanguageConfig
                                                                              audioConfiguration:audioConfig];

In the following example, the source language and custom endpoint are provided by using SPXSourceLanguageConfiguration. Then, SPXSourceLanguageConfiguration is passed as a parameter to the SPXSpeechRecognizer construct.

SPXSourceLanguageConfiguration* sourceLanguageConfig = \
        [[SPXSourceLanguageConfiguration alloc]initWithLanguage:@"de-DE"
                                                     endpointId:@"The Endpoint ID for your custom model."];
SPXSpeechRecognizer* speechRecognizer = [[SPXSpeechRecognizer alloc] initWithSpeechConfiguration:speechConfig
                                                                     sourceLanguageConfiguration:sourceLanguageConfig
                                                                              audioConfiguration:audioConfig];

Note

The speechRecognitionLanguage and endpointId properties are deprecated from the SPXSpeechConfiguration class in Objective-C. The use of these properties is discouraged. Don't use them when you create a SPXSpeechRecognizer construct.

Next steps