Decided on Parakeet ONNX Runtime. Works pretty great. Realtime voice chat possible now. UX lacking.

stt-parakeet/example.py

@@ -0,0 +1,15 @@
+"""
+Simple example of using the ASR pipeline
+"""
+from asr.asr_pipeline import ASRPipeline
+
+# Initialize pipeline (will download model on first run)
+print("Loading ASR model...")
+pipeline = ASRPipeline()
+
+# Transcribe a WAV file
+print("\nTranscribing audio...")
+text = pipeline.transcribe("test.wav")
+
+print("\nTranscription:")
+print(text)