Transcribe Audio to Text — Free AI Transcription in 99 Languages

VexaScribe (formerly NovaScribe) uses OpenAI's Whisper Large-v3 model to convert speech to text. Upload an audio or video file, and the AI processes the entire recording — adding speaker labels, word-level timestamps, and optional AI summaries. A 60-minute file typically completes in 5-10 minutes.

VexaScribe accepts MP3, WAV, M4A, FLAC, OGG, AAC, AIFF, WMA, AMR, OPUS for audio, and MP4, MOV, AVI, MKV, WebM, FLV, WMV for video. Files can be up to 5 GB and 10 hours long. For video files, we extract the audio track automatically.

Most 1-hour files complete in 5-10 minutes. Processing speed depends on audio quality, current load, and file format. You can close the browser tab and return — the transcript will be waiting in your dashboard when it's ready.

Yes, you get 30 minutes of transcription free with no credit card required. After the free tier, paid plans start at $2/month for 200 minutes (Starter), $5/month for 1,000 minutes (Basic), $10/month for 2,500 minutes (Pro), and $20/month for 6,000 minutes (Studio). Cancel anytime.

VexaScribe achieves around 95% accuracy (5% Word Error Rate) on clear English audio with a single speaker. Real-world accuracy varies: clear podcast audio averages 3-6% WER, noisy interviews 8-15% WER, and audio with strong accents or technical jargon 10-20% WER. We recommend reviewing transcripts before publishing critical content.

99 languages including English, Spanish, French, German, Italian, Portuguese, Dutch, Russian, Polish, Japanese, Chinese, Korean, Arabic, Turkish, Hindi, Vietnamese, Thai, and many more. Language is detected automatically — no need to select it manually before each upload.

Yes. Upload MP4, MOV, AVI, MKV, WebM, FLV, or WMV files and we extract the audio track automatically. The transcript includes timestamps so you can sync with your video editing tool, generate subtitles (SRT/VTT export), or repurpose video content into blog posts.

Yes, automatic speaker diarization is included on every transcript. Multiple speakers are labeled Speaker 1, Speaker 2, Speaker 3, and so on. You can rename speakers in the built-in editor (e.g., "Host", "Guest", actual names) for clarity in the final transcript.

Audio files transit over TLS 1.2+ encryption and are stored encrypted at rest in AWS eu-west-2. We do not train AI models on your audio. We do not sell user data. You can delete files at any time from your dashboard, and account deletion is self-serve.

VexaScribe exports to TXT (plain text), DOCX (Word document), SRT (video subtitles), VTT (web subtitles), and JSON (structured data for developers). All formats are available on every paid plan. SRT and VTT include word-level timestamps for video editors.

Sign up at VexaScribe (30 minutes free, no credit card), drag your recording onto the upload area, and wait — the audio-to-text transcription completes in roughly 8–12% of the file's duration. A 60-minute recording is ready in 5–10 minutes. The output includes speaker labels, word-level timestamps, and an editable transcript view, and you can export the audio to text transcript as TXT, DOCX, SRT, VTT, or JSON. Works for voice memos, interviews, podcasts, lectures, Zoom calls, and phone recordings.

Yes. Every new account gets 30 minutes of free audio transcription to text with all features enabled — speaker diarization, 99-language support, timestamps, and export to TXT, DOCX, SRT, VTT, and JSON. No credit card required to start. After the free tier, paid plans start at $2/month for 200 minutes (about $0.01 per minute), which is significantly cheaper than typical pay-per-minute transcription services charging $0.10–$0.25 per minute.

A bare audio-to-text converter usually returns a wall of raw text with no speakers, no timestamps, and no editor — you have to clean it up yourself. A transcription service like VexaScribe returns a structured transcript: speakers are labeled (Speaker 1, Speaker 2…), every word is timestamped to the second, the text is paragraph-broken for readability, you can edit and re-export in-browser, and AI summaries with action items are generated automatically on paid plans. Same upload, much more usable output.

Three steps, converter-style. (1) Pick your audio file — MP3, WAV, M4A, FLAC, OGG, AAC, AIFF, WMA, AMR, or OPUS, up to 5 GB per file — and drag it into VexaScribe, or paste a Google Drive share link / direct audio URL and skip the upload. (2) The AI converts speech to text in 5-10 minutes for a 1-hour file, with speaker labels and timestamps added automatically. (3) Download the text as TXT, DOCX, SRT, VTT, or JSON. Unlike single-format converter sites, one upload produces all five output formats — no re-processing per format.

All ten common audio containers: MP3 (the most common), WAV (uncompressed studio/voice-recorder), M4A (iPhone voice memos, AAC), FLAC (lossless), OGG (WhatsApp voice notes use Opus-in-OGG), AAC, AIFF (Mac audio), WMA (legacy Windows), AMR (old phone recordings), and OPUS. Video containers work too — MP4, MOV, AVI, MKV, WebM, FLV, WMV — with the audio track extracted automatically. If you have a format-specific question, see our dedicated guides: MP3 to text, WAV to text, M4A to text, OGG to text, MP4 to text.

Yes — VexaScribe converts your first 30 minutes free with no credit card, including speaker labels and all export formats. That's enough for most one-off conversions. Other free options have tighter catches: Canva's converter caps files at 4.5 MB (about 4 minutes of MP3), TurboScribe's free tier allows 3 files/day at 30 minutes each, and Otter's Basic tier caps recordings at 30 minutes with limited file imports. For unlimited free conversion, self-hosted OpenAI Whisper costs nothing but requires Python and a GPU. See our full free transcription comparison for every option's real limit.

Open the Voice Memos app on your iPhone. Tap the voice memo you want to transcribe. Tap the three-dot menu, then Share. Choose Save to Files (or AirDrop / Mail / Files). The recording saves as an .m4a. Drop the .m4a into VexaScribe here — the transcript comes back in about 5-15 seconds for a typical 1-minute memo, complete with timestamps and paragraph breaks. iOS 18 added native voice-memo transcription, but it's English-only for the polished view and doesn't produce a portable text file for export. Our tool works in 99 languages, produces a portable transcript in TXT/DOCX/JSON, and handles memos of any length up to 4 hours.

iOS 18 added on-device voice-memo transcription — English only for the polished experience, viewable in the Voice Memos app but not directly exportable as a text file. Use ours when: (1) the memo is in a language other than English (Turkish, Spanish, Mandarin, Arabic, and 95 others); (2) you want a clean .txt or .docx to paste into a note-taking app or CMS; (3) you want speaker labels for multi-person recordings (iOS doesn't diarize); (4) you're on iOS 17 or older; (5) accuracy matters — Whisper Large-v3 typically outperforms Apple's on-device model by 5-10 percentage points on accented speech and noisy audio.

Yes. Zoom cloud recordings download as .mp4 (audio+video) or .m4a (audio-only); drop either into VexaScribe. Google Meet cloud recordings download as .mp4. Microsoft Teams recordings download as .mp4 from the meeting chat. All work identically — we extract the audio track from the video container and run Whisper with speaker diarization. For a typical 60-minute meeting with 4-6 speakers, the transcript is ready in 5-10 minutes and each speaker is tagged Speaker 1, Speaker 2, etc. Rename them to real names in the editor and the change applies across the whole transcript. For live meeting transcription (during the call, not after), see /meeting-transcription — that page uses a meeting bot that joins the call.

Yes for one-off transcription of a specific episode. Drop the .mp3 (from your podcast host's download link, or from Apple/Spotify/Google Podcasts if you exported the audio) and the transcript comes back with host/guest speaker labels. For an ongoing podcast workflow — auto-transcribing every new episode as it publishes — /podcast-transcription covers the RSS integration and per-episode automation. Both use the same Whisper Large-v3 engine and produce the same output quality. A typical 60-minute podcast episode with two speakers hits 93-96% accuracy after 5-10 minutes of processing.

Yes — this is one of the highest-accuracy use cases for the tool. Two-speaker interviews (interviewer + subject) with clean mics hit 94-97% accuracy. Speaker diarization automatically labels each turn Speaker 1 / Speaker 2 — rename to real names in the editor. For qualitative research interviews requiring verbatim transcription and coding, see /transcription-for-qualitative-research (same engine, extra workflow guidance for NVivo/ATLAS.ti/MAXQDA import). For legal depositions or news journalism where verbatim accuracy is required for direct quoting, budget 10-15 minutes of review per interview hour to fix proper nouns and confirm speaker attribution.

Same Whisper Large-v3 engine, different UX and target user. This page (/transcribe-audio-to-text) is the general audio-to-text tool — voice memos, meetings, podcasts, interviews, any audio file. /voicemail-to-text is specialized for phone voicemails: it includes per-carrier export instructions (iPhone Visual Voicemail, Android voicemail apps, Google Voice, business VoIP), an accuracy comparison against carrier transcription, and short-form-audio-optimized UX. If your audio is a voicemail, /voicemail-to-text has better instructions. If it's a voice memo, meeting, podcast, or any other audio, this page is faster.

Three steps. (1) Upload your file — MP3, WAV, M4A, OPUS, OGG, FLAC, AAC, AIFF, WMA, or AMR, up to 5 GB and 10 hours per file. Or paste a Google Drive share link. (2) Whisper Large-v3 processes the audio in about 8-12% of the recording's length — a 60-minute file returns a transcript in 5-10 minutes. Speaker diarization, timestamps, and paragraph breaks are added automatically. (3) Download the transcript object as TXT, DOCX, SRT, VTT, or JSON. The transcript stays in your dashboard for re-download or export in additional formats without re-processing. 30 minutes free, no credit card required to start.

Framing, not technology — same speech recognition underneath. An 'audio-to-text converter' typically refers to a quick single-format utility: paste or upload audio, get a text blob back, done. A 'transcript generator' produces a structured transcript object: speaker labels (Speaker 1, Speaker 2), per-word timestamps synced to the audio, paragraph breaks by pause or speaker change, and an editable in-browser view before you download. VexaScribe produces the transcript object by default. Every plan — including the free 30-minute trial — exports in TXT, DOCX, SRT, VTT, and JSON from a single upload, so you don't re-process to switch formats.

Same 3 steps as any audio file. (1) Save the voice recording as a file on your computer — from a phone recorder app, dedicated audio recorder (Sony ICD, Zoom H1n, Tascam DR series), or browser-based recorder. (2) Drop the file (MP3, M4A, WAV, OPUS, OGG, FLAC — up to 5 GB, 10 hours) into VexaScribe. (3) Wait 5-10 minutes per audio hour — Whisper Large-v3 runs at 4-10× real-time. Download the transcript in TXT, DOCX, SRT, VTT, or JSON. Works on recordings from any device, in 99 languages, with speaker labels and word-level timestamps. First 30 minutes free, no credit card.

Depends on your priorities. For publication-quality quotes: Rev (human) at 99%+ accuracy, $1.99/min. For best AI value: VexaScribe at 93-95% Whisper Large-v3, $2-20/month, EU hosting. For meeting-note UX: Otter at ~88% English, $16.99/mo Pro. For transcript-first editing: Descript at 95%, $24-45/mo. For EU privacy + human option: HappyScribe at 92% AI or 99% human, €15-72/mo. For developer API: Deepgram Nova-3 or AssemblyAI Universal-2 at $0.0043-0.006/min. The honest cheap default for one-off recordings is VexaScribe's 30-min free trial; for ongoing needs, pick by budget and privacy requirement.

Three legitimate free paths. (1) VexaScribe 30-minute free trial — one-time, no credit card, all export formats (TXT/DOCX/SRT/VTT/JSON), speaker labels, 99 languages. Enough for most one-off recordings. (2) OpenAI Whisper self-hosted — free forever with Python + a GPU (or slow CPU). Best privacy since nothing leaves your machine. (3) Browser-based free tools with tighter limits — TurboScribe free tier (3 files/day), HappyScribe (10 min free), Otter Basic (30 min recording limit). For continuous free use, self-hosted Whisper is the only option; for a one-time transcription, the VexaScribe trial covers a typical 25-30 minute recording end-to-end.

Two paths. Web-based (recommended): drop the recording file into any browser-based transcription tool (VexaScribe, Otter web, Rev web, HappyScribe web). Works on any operating system — Windows, macOS, Linux, ChromeOS. No installation, output ready in 5-10 minutes for a 60-minute recording. Desktop app (offline): install OpenAI Whisper via Python and run locally, or use whisper.cpp (C++ port) which runs on CPU without Python. Requires more setup but produces transcripts without the audio ever leaving your machine — best for sensitive content. VexaScribe's web workflow covers most needs; whisper.cpp is the honest recommendation when privacy is paramount.

Two different jobs. Voice recording software captures audio from a microphone (Windows Voice Recorder, Audacity, Voice Memos on iPhone/Mac, Samsung Voice Recorder on Android). It saves an audio file — but doesn't convert it to text. Transcription software takes an existing audio file and produces text (VexaScribe, Rev, Otter, Whisper). Modern all-in-one tools do both — Otter records + transcribes live, Descript imports + transcribes. For a workflow where you already record on your phone and want text later, use any recording app + a dedicated transcription tool (upload the file). For live meeting note-taking, an all-in-one bot (Otter, Fireflies, Fathom) or dedicated meeting-note-taker page is a better fit.