VOICE TO TEXT

Most free voice-to-text tools use the browser's built-in Web Speech API. That means Chrome-only, no speaker labels, no export formats beyond plain text, and inconsistent accuracy across devices. The paid alternatives (Otter, Notta) lock speaker diarization and multi-format export behind subscription tiers and require signup.

We use server-side Gemini 3.1 Flash Lite — the same AI model that powers YouTube auto-captions — so accuracy and features are consistent on every browser. Plus we ship live-mic recording AND file upload, speaker labels, and six export formats all on one free page.

What Makes Us Different

• Live mic + audio upload in one tool. Speechnotes is mic-only. Otter is upload-heavy. We give you both on the same page — pick your input, same output pipeline.
• Works in ALL browsers. Web Speech API tools (Dictation.io, SpeechTexter, voicetotextonline.com) only work in Chrome. We use server-side Gemini — Safari, Firefox, Edge, mobile browsers all get identical quality.
• Speaker diarization on the free tier. Otter gates this. Notta gates this. We include it by default — great for interview clips, short meetings, couple conversations.
• 6 export formats in one click — TXT (plain), TXT with timestamps + speaker labels, SRT (for video editors), VTT (web captions), JSON (structured for developers), Markdown with clickable timestamps (unique).
• Live browser recording with MediaRecorder API — waveform level meter, pause/resume, automatic stop at 5 min, cross-browser compatible mime-type selection.
• Embedded audio player + click-to-jump. Click any timestamp in the transcript, the recorded/uploaded audio jumps to that second for instant verification.
• 40+ language cross-lingual output. Dictate in English, export in Spanish. Transcribe a Japanese voice memo in English. Same pass, same accuracy.
• Per-segment hover copy. Hover any line, click copy — grab just that quote without manually selecting.
• No signup, no email, no credit card. 3 free runs per browser session. Pro upgrade removes the cap.

Who This Is For

• Journalists — transcribe interview voice memos without a subscription; get speaker-labelled quotes in seconds.
• Podcasters — dictate show-note drafts or pull quotes from short segments.
• Writers and bloggers — dictate a 5-minute chunk instead of typing; export as Markdown ready to paste.
• Students — voice-note a study summary, lecture observation, or research idea and get it instantly text-ready.
• Coaches and therapists — capture session reflections. Speaker labels help when 2 voices appear in the recording.
• Lawyers and consultants — transcribe short client voice memos without sending audio to a subscription service.
• Anyone on Safari or Firefox — who's tired of being told “open this in Chrome” by Web Speech tools.

Technical Notes

Mic input is captured via the MediaRecorder API with opus-encoded WebM (or platform-best codec fallback). An analyser node drives the live waveform meter. Recording auto-stops at the 5-minute mark. Uploaded audio is validated client-side (size + duration via HTMLAudioElement metadata) before the request fires. Transcription is powered by Google Gemini 3.1 Flash Lite Preview through OpenRouter in JSON mode. Timestamps preserve sub-second precision. Speaker labels come from voice-based turn detection — not a pre-defined roster.