VoiceBlender v0.2.0 is now available — packed with new capabilities for voice application developers.
What’s New
Answering Machine Detection (AMD)
One of the most requested features for outbound calling is here. VoiceBlender can now detect whether a call was answered by a human or a machine, including beep detection. This allows your application to make smarter decisions — leave a voicemail when a machine picks up, or connect a live agent when a human answers.
Azure Speech Services Support
In addition to ElevenLabs, Google Cloud, AWS Polly, and Deepgram, VoiceBlender now supports Microsoft Azure Speech Services for both text-to-speech and speech-to-text. This gives teams already invested in the Azure ecosystem a native integration path.
Speech Detection Rework
The speech detection pipeline has been reworked for improved accuracy and lower latency. This benefits AI agent conversations, barge-in behavior, and any workflow that depends on detecting when a caller starts or stops speaking.
Other Improvements
- Added test files for better coverage
- Small quality-of-life improvements across the codebase
Upgrading
Pull the latest release from GitHub:
go get github.com/VoiceBlender/voiceblender@v0.2.0
go build -o voiceblender ./cmd/voiceblender
Full changelog: v0.1.0…v0.2.0
Check the API Documentation for details on the new AMD and Azure endpoints.