armenian-medical-speech-recognition
tigran0005/armenian-medical-speech-recognition
armenian-speech-recognitionmedical-nlpwhisper-asrhugging-face-transformerspytorchwer-cer-evaluationspeech-processingjupyter-notebook
Summary
An experimental Armenian medical speech recognition project using a pretrained Whisper model via Hugging Face Transformers. It focuses on transcription of Armenian medical audio, particularly traumatology, with chunk-based processing and evaluation using WER and CER. The results are based on a single 47-second recording and are preliminary.