BETASDK v1.1.0
Real-time speech-to-text and voice workflows.
The official Python client for the UltraSafe AI ASR API. Transcribe audio, stream in real time, and translate speech — with a small, strictly-typed SDK that does one thing well.
Need an API key? Email us and we'll set you up — keys are issued manually while we're in beta.
Quickstart
Install the wheel, export your key, transcribe a file. Three commands.
# 1. Install
pip install \
https://d3a9v9y9w3meag.cloudfront.net/wheels/1.1.0/ultrasafe_asr-1.1.0-py3-none-any.whl
# 2. Configure
export USF_API_KEY="<the key we sent you>"
export USF_BASE_URL="https://api-prod-usf.us.inc"
# 3. Transcribe
python -c "from ultrasafe_asr import UsfClient; \
print(UsfClient().transcribe('audio.wav').text)"Everything you need, nothing you don't
A small, auditable surface that covers file transcription, real-time streaming, and translation — with VAD and diarization configurable per request.
File & real-time transcription
Transcribe WAV/MP3/FLAC files or stream microphone audio over WebSockets with a single, fully-typed client.
Built for production
Small, focused codebase with strict typing, clean exception hierarchy, and first-class async support.
Never leaks your master key
Your testers get a per-user API key; the SDK talks to a proxy that substitutes in the real upstream credential.
VAD + diarization
Opt-in voice activity detection and speaker diarization — configurable per request. Standalone audio enhancement is on the roadmap.
Context-managed resources
Both sync and async clients support `with` blocks; no leaked file handles, no dangling WebSockets.
OpenAI-compatible
Drop-in replacement for OpenAI Whisper API patterns; the SDK surfaces are familiar and predictable.
Ready to start?
Download the SDK and start transcribing in under 60 seconds. Browse the full docs for real-time streaming, VAD, and diarization.