mikael-lovqvists-claude-agent
18404708e3
Every connection receives the full event stream (vad_start, vad_end, transcript, error) from the moment it connects — no subscription handshake required. The asyncio WebSocket server runs in a daemon thread alongside the VAD loop and transcription thread. Events still go to stdout unchanged. Port is configurable via STT_PORT env var (default: 11501). Add websockets to both setup scripts. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
32 lines
1.0 KiB
Bash
Executable File
32 lines
1.0 KiB
Bash
Executable File
#!/usr/bin/env bash
|
|
#
|
|
# setup-venv.sh — installs faster-whisper from PyPI into a venv.
|
|
#
|
|
# USE THIS SCRIPT when the PyPI ctranslate2 wheel matches your CUDA version.
|
|
# PyPI wheels target a specific CUDA major version; if your system matches,
|
|
# this is the fastest way to get started — no compilation required.
|
|
#
|
|
# If you see errors like "libcublas.so.12: cannot open shared object file" at
|
|
# runtime, your CUDA version does not match the wheel. Use setup-venv-local-build.sh
|
|
# instead, which compiles ctranslate2 against your actual CUDA installation.
|
|
#
|
|
# Environment overrides:
|
|
# PYTHON_ENV path to venv (default: ./venv)
|
|
#
|
|
set -euo pipefail
|
|
|
|
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
|
|
VENV="${PYTHON_ENV:-${SCRIPT_DIR}/venv}"
|
|
|
|
if [ ! -d "${VENV}" ]; then
|
|
echo "==> creating venv at ${VENV}"
|
|
python3 -m venv "${VENV}"
|
|
fi
|
|
|
|
echo "==> installing torch and faster-whisper"
|
|
"${VENV}/bin/pip" install --upgrade pip --quiet
|
|
"${VENV}/bin/pip" install torch faster-whisper silero-vad websockets
|
|
|
|
echo ""
|
|
echo "==> done. Venv ready at ${VENV}"
|