The State of Voice Interactions on Android
summary
Timeline shows Android speech APIs since 2009–2010 with on‑device recognition added for Android 13 (2020) and language detection in 2023, and practical stacks combine wake‑word detection, speech‑to‑text, NLP/intent parsing (Dialogflow or LLMs like Gemini), business logic, and text‑to‑speech. Demos use Whisper, Meta Wit.ai, PicoVoice and Hugging Face models; Gemini API accepts audio and Gemini Nano or MediaPipe enable local LLM inference. Pick only the blocks you actually need.
AI-written summary · about
chapters (21)
- 0:00Voice interactions on Android
- 1:10Why voice matters today
- 2:58Four blocks for voice apps
- 6:10LLMs in voice experiences
- 10:33Three voice integration strategies
- 12:16Android APIs for voice
- 16:57Voice Access feature
- 17:18Google Assistant integration
- 21:13Gemini voice capabilities
- 24:00On-device LLM with MediaPipe
- 26:52Third-party voice solutions
- 27:57Pico Voice offline platform
- 29:03Amazon & Microsoft voice APIs
- 29:45OpenAI Whisper STT
- 30:49Voyage NLP solutions
- 31:35Meta’s Wit AI for voice
- 32:37Custom voice models with Hugging Face
- 35:25Voice apps use cases
- 36:52Building voice experiences
- 38:57Future of voice on Android
- 40:45Q&A
related talks
42mThe State of Voice Interactions on Android
47mBuilding Voice-First Android Apps- 47mBuilding Voice-First Android Apps
29mAndroid Gets a Voice: Building Real-Time Agents
35mHaving a Voice on Android: How hard can it be?
55mConversational power for Android apps
16mHow to voicify your Android app
40mConversational power for Android apps