File size: 582 Bytes
c2ac364
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
# requirements.txt
fastapi
uvicorn[standard]
websockets
openai-whisper
torch
torchaudio
transformers
accelerate # Often useful for transformers
python-multipart # For file uploads in traditional endpoints
soundfile # For handling audio files
librosa
parler-tts # For AI4Bharat's IndicParler-TTS
onnx
onnxruntime
# For specific hardware acceleration (optional, choose based on your setup)
# bitsandbytes # For 8-bit quantization of LLM (further RAM reduction)
# sentencepiece # Often a dependency for tokenizers
# For demo purposes
gradio
# For Gemini integration
google-generativeai