MariusMyburg / UEAzSpeech

Plugin: AzSpeech

The project:

A plugin integrating Azure Speech Cognitive Services to Unreal Engine with only 2 simple functions which can do these asynchronous tasks:

1. TextToVoiceAsync
2. VoiceToTextAsync

Product on Unreal Engine Marketplace: https://www.unrealengine.com/marketplace/en-US/product/azspeech-async-text-to-voice-and-voice-to-text
Post on Unreal Engine Forum: https://forums.unrealengine.com/t/free-azspeech-plugin-async-text-to-voice-and-voice-to-text-with-microsoft-azure/495394

Documentation

Installation

Just clone this repository or download the .zip file in the latest version and drop the files inside the 'Plugins/' folder of your project folder. Note that if your project doesn't have a 'Plugins' folder, you can create one.

Blueprint Usage:

You have these asynchronous functions to manage all the workaround:

1. Text to Voice Async: Will convert the specified text to a sound that will be played by the default sound output device.
   1.1. Text to Convert: The text that will be converted to audio;
   1.2. Voice Name: Voice code that will represent the type of voice that will "read" the converted text. You can see all names here: https://docs.microsoft.com/en-us/azure/cognitive-services/speech-service/language-support#text-to-speech;
   1.3. Parameters: Microsoft Azure parameters to connect to the service and perform the tasks. The structure AzSpeechData represents this input;

2. Voice to Text Async: Will convert your speech to string. Not that this function will get the speech by your default sound input device.
   2.1. Parameters: Microsoft Azure parameters to connect to the service and perform the tasks. The structure AzSpeechData represents this input;

3. AzSpeechData: Represents Microsoft Azure parameters to connect to the service and perform the tasks;
   3.1. Subscription ID: It's your Speech Service Subscription ID from your Microsoft Azure Portal;
   3.2. Region ID: Region that will be passed to Speech service. You can see all regions here: https://docs.microsoft.com/en-us/azure/cognitive-services/speech-service/regions;
   3.3. Language ID: Language to apply lozalization settings. You can see all IDs here: https://docs.microsoft.com/en-us/azure/cognitive-services/speech-service/language-support#speech-to-text;

C++ Implementation (Text-to-Voice):

(Using GameMode class only for demonstration)

You need to include the module "AzSpeech" inside your .Build.cs class to be allowed to call AzSpeech functions from C++.

1. UTextToVoiceAsync::TextToVoiceAsync(WorldContextObject, TextToConvert, VoiceName, FAzSpeechData): Will convert the specified text to a sound that will be played by the default sound output device.
   1.1. WorldContextObject: Task Owner.
   1.2. TextToConvert: The text that will be converted to audio;
   1.3. VoiceName: Voice code that will represent the type of voice that will "read" the converted text. You can see all names here: https://docs.microsoft.com/en-us/azure/cognitive-services/speech-service/language-support#text-to-speech;
   1.4. FAzSpeechData: Microsoft Azure parameters to connect to the service and perform the tasks. The structure AzSpeechData represents this input;

2. FAzSpeechData: Represents Microsoft Azure parameters to connect to the service and perform the tasks;
   2.1. Subscription ID: It's your Speech Service Subscription ID from your Microsoft Azure Portal;
   2.2. Region ID: Region that will be passed to Speech service. You can see all regions here: https://docs.microsoft.com/en-us/azure/cognitive-services/speech-service/regions;
   2.3. Language ID: Language to apply lozalization settings. You can see all IDs here: https://docs.microsoft.com/en-us/azure/cognitive-services/speech-service/language-support#speech-to-text;

The TextToVoiceAsync function return a delegate to manage task status which allow you to bind a function to it's delegate to handle task's completion call.

C++ Implementation (Voice-to-Text):

(Using GameMode class only for demonstration)

You need to include the module "AzSpeech" inside your .Build.cs class to be allowed to call AzSpeech functions from C++.

1. UVoiceToTextAsync::VoiceToTextAsync(WorldContextObject, FAzSpeechData): Will convert the specified text to a sound that will be played by the default sound output device.
   1.1. WorldContextObject: Task Owner.
   1.2. FAzSpeechData: Microsoft Azure parameters to connect to the service and perform the tasks. The structure AzSpeechData represents this input;

2. FAzSpeechData: Represents Microsoft Azure parameters to connect to the service and perform the tasks;
   2.1. Subscription ID: It's your Speech Service Subscription ID from your Microsoft Azure Portal;
   2.2. Region ID: Region that will be passed to Speech service. You can see all regions here: https://docs.microsoft.com/en-us/azure/cognitive-services/speech-service/regions;
   2.3. Language ID: Language to apply lozalization settings. You can see all IDs here: https://docs.microsoft.com/en-us/azure/cognitive-services/speech-service/language-support#speech-to-text;

The VoiceToTextAsync function return a delegate to manage task status which allow you to bind a function to it's delegate to handle task's completion call.

More informations:

See the official Microsoft documentation for Azure Cognitive Services: https://docs.microsoft.com/en-us/azure/cognitive-services/speech-service/