Auzure Speech To Text impl.

[stone-alex]

Description

Add Microsoft Azure Cognitive Services Speech-to-Text as a second STT backend.
The app will automatically detect the provider from the API key format (no user dropdown needed).

Why

• Gives players choice between Google and Azure
• Zero config for existing users
• Fully backwards compatible

Acceptance criteria

• Implement elite.intel.ai.ears.EarsInterface in new class AzureSTT
• Must be singleton, zero DI – follow exact pattern of GoogleSTT
• Loaded dynamically via elite.intel.ai.ApiFactory using elite.intel.ai.KeyDetector.detectProvider(apiKey, "STT")
• Reuse or adopt existing RMS gate, VAD, and Audio Calibration logic (see from GoogleSTT)
• Use existing EventBusManager.register(this); / @subscribe pattern to listen to events
• Interrupt the STT procerssing when vocalization is in progress (See GoogleSTT / IsSpeakingEvent) and resume after vocalization ends.
• No Python, no JNI, no unsigned DLLs no native packages or libraries → pure Azure Java SDK only
• Fully works on Linux and Windows
• Project compiles and runs with gradle
• App runs from fat-jar
• Provide temporary key to project maintainer for testing.

Useful links

• Azure Speech SDK (Maven Central): https://central.sonatype.com/artifact/com.microsoft.cognitiveservices.speech/client-sdk
• Continuous recognition sample: https://github.com/Azure-Samples/cognitive-services-speech-sdk/blob/master/samples/java/jre/console/src/com/microsoft/cognitiveservices/speech/samples/console/SpeechRecognitionSamples.java

Difficulty: Medium (follow GoogleSTT pattern + async streams)