Speech recognition, speech synthesis, speaker diarization, and speaker recognition using next-gen Kaldi with onnxruntime without Internet connection.