Building a voice-follow teleprompter on iPhone without sending audio anywhere
A teleprompter that scrolls as you speak sounds simple: recognize speech, find the words in the...
Tag archive
A teleprompter that scrolls as you speak sounds simple: recognize speech, find the words in the...

ClinicLog turns a recorded therapy session in Hebrew, English or both into a transcript and a draft note the therapist signs. Models tidy the transcript; plain code decides which of their changes are kept.

Inclusive teaching and learning with real-time 9-language live subtitles handling Hong Kong English-Cantonese code-switching via Gemini 4 and Gemma 4 E2B.
Azure’s Oct 2026 LLM Speech update adds 25 languages (11 Indic) and up to 2× speed on short audio. Practical Fast Transcription notes for Windows tools.

A WhatsApp Voice Note Is Input, Not Permission to Act A voice note can contain a request,...

Hugging Face ships Swift packages for pluggable on-device speech recognition and typed decisions. Architecture notes for real-time audio tools.

Ollama 0.40.13 brings Liquid LFM3 audio models to Windows. What a 2B local STT/TTS stack actually means for live call capture, latency, and hybrid pipelines.

Measurements of Russian STT on a CPU-only VM: T-one, GigaAM-v3 and Gemini models on my own recordings, why I didn't report WER, plus short notes on TTS, VAD and licenses.

Built an all-local streaming JA-EN speech translator and measured the verb-final wait cost: EN-JA prefix translations survive ~13x worse than JA-EN at matched prefixes (same backend).
Typing is slow for a lot of people, and not just because of speed. Some people think out loud far...

Block's Dessa was trained only on recorded human conversation, no text, and runs on a MacBook Pro M4 Max. Safety researchers get it first; scam fears follow.
A research proposal by Muhammad Usman, BS Artificial Intelligence, Superior University, Lahore —...