# MrrrMe Backend Requirements - Coqui XTTS v2 + emotion2vec # Core frameworks fastapi==0.115.4 uvicorn[standard]==0.32.0 python-multipart==0.0.12 websockets==13.1 # Computer Vision opencv-python-headless==4.10.0.84 pillow==11.0.0 scikit-image==0.24.0 matplotlib==3.9.2 timm==1.0.11 einops==0.8.0 # ML/DL # NOTE: torch, torchvision, torchaudio are installed in Dockerfile numpy==1.26.4 scipy==1.13.1 pandas==2.2.3 # Unpinned accelerate accelerate sentencepiece==0.2.0 safetensors==0.4.5 # Audio processing librosa==0.10.2.post1 soundfile==0.12.1 pyaudio==0.2.14 pydub==0.25.1 sounddevice==0.5.1 webrtcvad==2.0.10 # Voice Emotion Recognition (emotion2vec+ Large - 93% accuracy) # CRITICAL: Install funasr without dependencies to avoid gradio hell # funasr>=1.0.0 (Moved to Dockerfile manual install) # Manually install funasr dependencies (skipping gradio) # modelscope (Moved to Dockerfile manual install) kaldiio hydra-core editdistance jieba torch-complex pytorch-wpe # Text-to-Speech (Qwen 3 TTS via transformers) qwen-tts # Transformers: Qwen TTS requires a very recent version (e.g. 4.57.3) transformers # accelerate is already listed above # Text processing nltk==3.9.1 # Groq API groq==0.14.0 # Utilities requests==2.32.3 python-dotenv==1.0.1 huggingface_hub>=0.20.0 tensorboardX==2.6.2.2