SAPI Bridge for Online TTS

This project demonstrates how to create a Python COM server that functions as a bridge between online Text to Speech services and other engines which dont have SAPI imterfaces like Microsoft Azure Speech SDK and the Speech Application Programming Interface (SAPI) used by Windows. It allows applications expecting SAPI compatibility to utilize Azure's advanced neural voices for text-to-speech (TTS) functionalities.

Quick notes:

• voiceBroker.py looks like it should work (use it with a admin shell). Use it with demoClient for example. This is the code we want to work from
• (voiceBroker-MSVanilla.py is not using tts-wrapper. Just for Azure, The code is a little older - i dare the creds might not work.
• look at basic_demo_register.py. Its a really basic version of this more complex code. Uing any of these com registration code in any SAPI system and it doesnt work. We need to debug this. I think somethings missing (e.g try it in https://www.cross-plus-a.com/balabolka.htm)
• Work with basic_demo_register.py and then find what is missing
• You need the correct credentials.json
• demoClientExtended - try it by all means but I have little faith in the onWord event handling being passed through correctly.

Here's what I think is going on. I think we have largely followed the details here. https://learn.microsoft.com/en-us/previous-versions/windows/desktop/ms720163(v=vs.85) We register the Engine and com service. That definitely works and we can get speak working with demoClient.py. But trying it any proper SAPI application doesnt. We are finding it hard to debug this. I have a feeling we have methods that arent implemented or speech isnt being called correctly. Im struggling to figure out from the MS docs whats going on.

ms docs: https://learn.microsoft.com/en-us/dotnet/api/system.speech.synthesis.ttsengine?view=net-8.0

Note: look at requirements. we are using py3.11.4 and using our own forks of py3-tts - you'll see why.

this is also some code that demonstrates how we could test it

   import win32com.client as wincl
   
   speaker_number = 1
   spk = wincl.Dispatch("SAPI.SpVoice")
   vcs = spk.GetVoices()
   SVSFlag = 11
   print(vcs.Item (speaker_number) .GetAttribute ("Name")) # speaker name
   spk.Voice
   spk.SetVoice(vcs.Item(speaker_number)) # set voice (see Windows Text-to-Speech settings)
   spk.Speak("Hello, it works!")

Aims

• Basic setting as a COM Service wih some functions
• Correctly register in the registry (although ids could be refactored)
• First work on Azure using tts-wrapper (why? Because I know we can get word level timing - it should help for callbacks)
• check other libraries
• Allow for configuration by a seperate app to allow for choice of voices etc. note how settings is the key file and a credentials file

Features

• Utilizes TTS-Wrapper to access a range of TTS systems and allow access via SAPI for high-quality neural voices.
• Implements SAPI-like methods such as Speak, Pause, Resume, and voice selection.
• Supports word-level timestamp events for synchronized operations.
• Designed to work with any SAPI-compatible software on Windows.

Prerequisites

Before you start using this project, ensure you have the following:

• Python 3.11.4 or higher
• Azure subscription and Speech service created on Azure portal. or polly etc
• Required Python packages: tts-wrapper (willwade fork), pythoncom, win32com.client, simpleaudio

Installation

1. Clone the Repository

   git clone https://github.com/AceCentre/VoiceBroker.git
   cd VoiceBroker

2, Install Dependencies

```bash
pip install -r requirements.txt
```

3. Configure Azure API Key and Region

Set up your Azure API key and region in the configuration section of the script or as environment variables.

Usage

To run the server and register it as a COM object on your Windows machine, execute:

python VoiceBroker.py --register

After registration, your Python COM server will be available for use by any SAPI-compatible software.

Methods Implemented

• Speak(text): Synthesizes speech from text using Azure's TTS.
• Pause(): Pauses the ongoing speech synthesis.
• Resume(): Resumes the paused speech synthesis.
• GetVoices(): Lists available voices from Azure.
• SetVoice(voice_name): Sets the current voice to the specified one.
• SetInterest(event_id, enabled): Subscribes or unsubscribes from specific speech events.
• WaitForNotifyEvent(timeout): Waits for a subscribed event to occur within a specified timeout period.

Contributing

Contributions are MORE than welcome. This code is not tested whatsoever. Please improve