InsuranceBot / requirements.txt
rohitsar567's picture
fix(stt): transcode browser webm/opus β†’ wav before Sarvam (D-024)
a777198
Raw
History Blame Contribute Delete
1.04 kB
# Web framework
fastapi==0.115.4
uvicorn[standard]==0.32.0
python-multipart==0.0.20
# Settings + types
python-dotenv==1.0.1
pydantic==2.9.2
# HTTP clients
httpx==0.27.2
requests==2.34.0
# PDF / document parsing
pdfplumber==0.11.4
# RAG: vector store + structured store
# 2026-05-14: pinned to 1.5.9 to match the local .venv version that builds
# rag/vectors. Schema drift between chroma versions caused KeyError('_type')
# in earlier deploys, wiping the index and forcing a 45-min re-ingest.
# Bump both local AND this pin together if upgrading.
chromadb==1.5.9
duckdb==1.1.3
# Local embeddings (BGE small) β€” replaces Voyage in v1 to bypass free-tier rate limits
sentence-transformers==5.1.2
torch==2.11.0 # CPU-only fine for our usage
# Audio transcoding β€” browser MediaRecorder emits webm/opus by default;
# Sarvam STT accepts WAV/MP3/FLAC/OGG/M4A but NOT webm. pydub (ffmpeg-backed)
# converts in-process. The Dockerfile installs the ffmpeg apt package.
pydub==0.25.1
# Utility
numpy==1.26.4
tqdm==4.66.6
tenacity==9.0.0