Crossword Diagnostics
Compare browser-native, OpenAI Realtime, and Google Cloud streaming transcription with timing diagnostics.
Uses the browser SpeechRecognition engine directly (no backend stream) for quick interim/final diagnostics.
Not listening
Browser stream: disconnected
Listening after start: n/a
First final loaded: n/a
No final browser transcript yet.
No interim browser transcript yet.
No completed browser utterances yet.
Uses OpenAI Realtime session + WebRTC stream + transcription delta events.
Cloud access requires sign-in and explicit project permission (`stt_lab`) for your account.
Not listening
Live session: disconnected
Status: Idle.
Listening after start: n/a
First final loaded: n/a
No final transcript yet.
No live partial transcript.
No completed utterances yet.
Uses Google Speech-to-Text v2 streaming (recognizer location is configurable) via/api/stt-lab/google-v2-stream.
Cloud access requires sign-in and explicit project permission (`stt_lab`) for your account.
Not listening
Google stream: disconnected
Status: Idle.
Listening after start: n/a
First final loaded: n/a
No final Google transcript yet.
No interim Google transcript yet.
No completed Google utterances yet.
No logs yet.