Spaces:
Runtime error
Runtime error
Commit History
-
Fix performance issues: updated requirements with exact versions and added model caching
Pranav Mishraa26c11a -
Update API version to 1.2.0 with ffmpeg support
Pranav Mishra +1: Claude- Version bump for ffmpeg system dependencies fix - Audio processing now has proper ffmpeg/ffprobe support 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>93ceaba -
Add ffmpeg system dependencies to fix audio processing
Pranav Mishra +1: Claude- Install ffmpeg and libsndfile1 in Dockerfile - Fix "No such file or directory: 'ffprobe'" errors - Enable proper audio format conversion for ML models 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>f3c42b1 -
Fix audio processing libraries for HF Spaces v1.1.0
Pranav Mishra +1: Claude- Add pydub, ffmpeg-python, audioread, resampy to requirements_hf.txt - Update API version to 1.1.0 - Fix audio corruption issues causing poor ML predictions 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>e46f847 -
Update README.md
Verified signaturefdb6ca6 -
Fix missing dependencies by simplifying dataset loader for inference
Pranav Mishra- Replaced full dataset_loader.py with inference-only version - Removed pandas and datasets dependencies (not needed for inference) - Kept only audio preprocessing functionality (preprocess_audio method) - Streamlined requirements.txt to remove heavy dependencies - Should resolve 'ModuleNotFoundError: No module named pandas' error - Smaller build size and faster deployment09061df -
Fix ML training module imports for HF Spaces deployment
Pranav Mishra- Added complete ml_training module to backend - Copied inference.py and all pipeline classes (MFCC, Mel CNN, Raw CNN) - Copied data loader module for audio preprocessing - Updated Dockerfile to include ml_training directory - Should resolve 'ModuleNotFoundError: No module named ml_training' errorb5ec43b -
Fix PyTorch compatibility issues for HF Spaces deployment
Pranav Mishra- Updated Dockerfile to use Python 3.10 instead of 3.9 for better package compatibility - Fixed requirements_hf.txt with working PyTorch versions (2.0.1/2.0.2) - Removed problematic CPU-specific PyTorch syntax - Updated numpy, scipy, and other packages to Python 3.10 compatible versions - Should resolve build errors and deploy successfully00672a5 -
Streamline backend: Remove Whisper/Wav2Vec2 models and dependencies
Pranav Mishra- Removed whisper_digit_processor.py, wav2vec2_processor.py, faster_whisper_processor.py - Removed local_whisper.py and related VAD utilities - Simplified requirements_hf.txt: removed transformers, webrtcvad, CPU-specific PyTorch - Only keeping 3 core ML models: MFCC, Mel CNN, Raw CNN + External API - Reduced build size and complexity for reliable HF Spaces deploymentdf60ee3 -
Initial backend deployment - Flask API with ML models
Pranav Mishra1772a46 -
initial commit
Verified signature494577d