Hugging Face Model Deployment and Library Dependency Issues

#16
by NeuraFusionAI - opened

https://github.com/neurafusionai/Hugging_Face/blob/main/audio/whisper_large_v3_german_asr(1).ipynb

Encountered challenges while deploying a fine-tuned Whisper model (openai/whisper-large-v3) for automatic speech recognition (ASR) on the Hugging Face Hub and working with the facebook/multilingual_librispeech dataset. The issues include:

Authentication Failures: Errors due to missing or incorrect Hugging Face API tokens when pushing the model and processor to the Hub.
Library Installation and Dependency Errors: Conflicts and missing packages, particularly related to libsndfile and other audio processing libraries, resulting in runtime errors when handling Opus files.
Model Path Issues: Errors due to incorrect paths when loading a locally fine-tuned model.
Inference Execution Errors: Challenges in correctly processing and transcribing audio data due to library dependencies and incorrect model path references.
The objective is to resolve these issues to successfully deploy and run the model for German speech recognition using the Whisper model and Hugging Face infrastructure.

Sign up or log in to comment