Learn how to detect language of an audio with Azure Speech API and Text Analytics

Learn how to detect language of an audio with Azure Speech API and Text Analytics

Table of Contents

• Introduction
• Phase 1: Extracting Text from an Audio File
  • Creating Speech Configuration Object
  • Creating Audio Configuration Object
  • Creating Language Configuration Object
  • Creating Recognizer Object
  • Extracting Text from Audio
• Phase 2: Detecting the Language
  • Installing Required Packages
  • Creating Azure Key Credential Object
  • Creating Text Analytics Client Object
  • Detecting the Language
• Conclusion

Introduction

In this video, we will learn how to extract text from an audio file and identify the language of the audio. The process will be divided into two phases. In the first phase, we will read the audio file and extract the text using the Speech API. In the Second phase, we will use Azure Text Analytics to detect the language of the audio. Let's get started!

Phase 1: Extracting Text from an Audio File

Creating Speech Configuration Object

To begin, we need to create a Speech Configuration object. This object will hold the subscription key and region. We can create the configuration object using the following code:

SpeechConfig speechConfig = SpeechConfig.FromSubscription("<subscription_key>", "<region>");

Creating Audio Configuration Object

Next, we need to create an Audio Configuration object to hold the audio file details. In our case, we are reading a web file. We can create the configuration object using the following code:

AudioConfig audioConfig = AudioConfig.FromWavFileInput("<file_path>");

Creating Language Configuration Object

Now, we need to create a Language Configuration object to specify the potential languages we want to identify. Let's say we want to identify English, Hindi, and Japanese languages. We can create the configuration object using the following code:

var languageConfig = LanguageDetectionConfig.FromLanguages(new List<STRING> { "en", "hi", "ja" });

Creating Recognizer Object

For Speech Recognition, we need to create a Recognizer object. We can create the object using the following code:

var recognizer = new SpeechRecognizer(speechConfig, audioConfig);

Extracting Text from Audio

We can now invoke the RecognizeOnceAsync method on the Recognizer object to extract the text from the audio file. Once the extraction is complete, we need to dispose of the resources. Here is an example of how to extract the text:

var result = await recognizer.RecognizeOnceAsync();
recognizer.Dispose();
return result.Text;

Phase 2: Detecting the Language

Installing Required Packages

To detect the language, we need to install the Azure.AI.TextAnalytics Package. We can do this using the package manager.

Creating Azure Key Credential Object

We need to create an Azure Key Credential object to hold the key for our Text Analytics resource. We can create the credential object using the following code:

AzureKeyCredential credential = new AzureKeyCredential("<api_key>");

Creating Text Analytics Client Object

Next, we can create a Text Analytics Client object using the endpoint and credential. Here is an example:

TextAnalyticsClient client = new TextAnalyticsClient(new Uri("<endpoint>"), credential);

Detecting the Language

Finally, we can use the client object to detect the language of the extracted text. Here is an example:

DocumentLanguage detectedLanguage = await client.DetectLanguageAsync("<text>");
Console.WriteLine("Detected Language: " + detectedLanguage.Name);

Conclusion

In this video, we learned how to extract text from an audio file and detect the language of the audio. We used the Speech API to extract the text and Azure Text Analytics to detect the language. This can be a useful tool for various applications that involve analyzing audio content.