DaVoice Blog
On-device voice AI insights for teams building private voice experiences
Technical articles, implementation guides, and product notes for wake word detection, hotword detection, speaker recognition, voice biometrics, speech to text, and text to speech across mobile, web, and embedded platforms.
Latest Post
Speaker Isolated Wake Word
All Posts
Technical writing for teams building private, low-latency voice AI.
Speaker Isolated Wake Word: Why Wake Word Detection Needs Speaker ID
Wake word detection answers whether a trigger phrase was spoken. Speaker identification answers whether it was spoken by the right person. Together, they make voice assistants safer and more reliable.
Read articleSpeaker Identified Speech to Text: Transcribing the Right Speaker
Speech to text tells a system what was said. Speaker identification tells it who said it. Together, they make voice systems personal, account-aware, safer, and more useful in noisy multi-speaker environments.
Read articleToken-Free Text to Speech with Near-Zero Latency
Cloud text to speech can create realistic voices, but every request adds latency, network dependency, and usage-based cost. DaVoice turns voices into compact on-device models that run locally without TTS tokens.
Read article