AccentPilot

AccentPilot is a real-time speech recognition overlay that improves caption accuracy for accented, fast, or noisy speech in video calls and in-person meetings. It runs as a desktop companion (Windows/macOS) that listens to the system audio/mic, generates low-latency captions, and lets users quickly correct misheard names, acronyms, and domain terms. Those corrections build a personal and team glossary that boosts accuracy over time without needing to retrain a full model. The app can output captions to a floating window, a browser source (for Zoom/Teams overlays), and export clean transcripts with speaker labels and action items. The realistic angle: it won’t beat native platform captions in every case, but it can win in edge cases—heavy accents, specialized vocabulary, and noisy rooms—where default captions regularly fail and users feel excluded.

← Back to idea list