Transcribe your first audio file in three steps.
Click "Choose File" or drag and drop an audio/video file onto the upload area. Supported formats include MP3, MP4, M4A, WAV, and WEBM. Files up to 100 MB are accepted.
Select your language (or leave on Auto-detect), enable Speaker Diarisation if you want speakers labelled separately, and click "Transcribe". Processing typically takes 10–30% of the audio duration.
Once complete, the transcript appears with speaker labels and timestamps. Rename speakers by clicking the speaker tag. Export to DOCX, SRT, or copy to clipboard.
Tip: For large files (> 50 MB) or files longer than 30 minutes, use the Storage URL path — upload via the app, then transcribe from the returned URL.
| Format | Extension | Notes |
|---|---|---|
| MP3 | .mp3 | Most common; recommended for voice recordings |
| M4A | .m4a | Default iPhone/iPad recording format |
| WAV | .wav | Uncompressed; highest quality but large files |
| WEBM | .webm | Browser-native recording format |
| MP4 | .mp4 | Video — audio track extracted automatically |
| MOV | .mov | Apple QuickTime — audio extracted |
| AAC | .aac | Advanced Audio Coding |
| Max file size | 100 MB |
|---|---|
| Direct upload limit | 4 MB* |
| Speakers | Up to 10 |
Note: * Files larger than 4 MB must be uploaded to Storage first (the app handles this automatically). The 4 MB limit applies only to direct base64 payloads — not to Storage-URL-based transcription.
LinguistPro is live for 13 languages — English and Afrikaans (SA-tuned) plus 11 global languages (Arabic, Chinese, French, German, Hindi, Italian, Japanese, Korean, Dutch, Portuguese, Spanish). Five South African indigenous languages (isiZulu, isiXhosa, Sesotho, Setswana, Tshivenda) are live in beta (best-effort, powered by Gemini). Leave on Auto-detect if unsure.
Powered by Gemini 2.5 Flash (short recordings) and AssemblyAI (large files), with Deepgram and Groq as fallbacks. Excellent accuracy for South African English accents, including Cape, Joburg, and KZN varieties. Supports diarisation up to 10 speakers.
Powered by Gemini 2.5 Flash (short recordings) and AssemblyAI Universal-2 (long recordings), with an automatic Dutch→Afrikaans correction pass so the verbatim reads as true Afrikaans rather than Dutch. Handles formal and informal Afrikaans, including English code-switching.
Live in beta, best-effort transcription powered by ElevenLabs Scribe (with Gemini fallback). Quality is still improving — we welcome your feedback.
Live in beta, best-effort transcription powered by ElevenLabs Scribe (with Gemini fallback). Quality is still improving — we welcome your feedback.
Live in beta, best-effort transcription powered by ElevenLabs Scribe (with Gemini fallback). Quality is still improving — we welcome your feedback.
These 11 global languages are live. Select one from the per-recording language picker, or set a default via Settings → Default language.
ar Arabic (العربية)zh Chinese (中文)fr French (Français)de German (Deutsch)hi Hindi (हिन्दी)it Italian (Italiano)ja Japanese (日本語)ko Korean (한국어)nl Dutch (Nederlands)pt Portuguese (Português)es Spanish (Español)Tip: Leave the language on Auto-detect for mixed-language recordings or if you're unsure. LinguistPro will identify the language automatically. Set a specific language only if Auto-detect gives incorrect results.
Diarisation identifies who spoke when and labels each segment with a speaker ID (e.g. Speaker 1, Speaker 2). LinguistPro adds an optional visual identification pass to assign real names from on-screen text.
After transcription, each speaker label becomes clickable. Click it to open the speaker naming panel. Type the person's real name and press Enter. All instances of that speaker update immediately across the transcript.
If your video recording contains visible name lower-thirds, overlays, or on-screen titles, LinguistPro's visual ID system attempts to match names to speakers automatically. This works best with Zoom, Teams, and Google Meet recordings where names are displayed consistently.
Tip: For best diarisation results, use a recording where speakers don't frequently talk over each other, and where each speaker has at least a few long turns.
Professional Word document with full speaker labels, timestamps, and optional AI intelligence sections (summary, action items, sentiment). Formatted for sharing and archiving.
Best for: meetings, interviews, legal records, formal reports
SubRip Subtitle format — the universal subtitle standard. Compatible with VLC, Premiere Pro, DaVinci Resolve, YouTube, and any other video editor or platform.
Best for: video captioning, accessibility, social media
Copies the plain text transcript to your clipboard with speaker labels intact. No formatting — ready to paste into any document, chat, or email.
Best for: quick sharing, pasting into other tools
R0 30 min / month
R7.50 per hour
R299 per month · coming soon
Click "Top Up" in the nav bar. Choose R50 (6h 40m), R100 (13h 20m), or R200 (26h 40m). Payment via PayFast — all major SA cards, EFT, and instant EFT supported. Credits are added to your balance within seconds of payment confirmation.
Free plan users receive 30 minutes of transcription on account creation. This allowance resets to 30 minutes on the 1st of every month — even if you haven't used all of it.
An unlimited monthly plan is coming soon and is not currently on sale.
Available to all signed-in users
After transcription completes, LinguistPro's AI post-processing layer (powered by Google Gemini) can generate additional intelligence from your transcript.
A concise, bullet-pointed summary of the main discussion points, key decisions made, and outcomes agreed upon. Typically 3–8 bullets for a one-hour meeting.
Extracts explicit commitments, tasks, and follow-ups mentioned in the recording. Each action item includes the responsible party where identifiable.
Tracks the emotional tone of the conversation over time. Identifies positive/negative/neutral shifts and flags sections with heightened tension or enthusiasm.
Analyses each speaker's communication style — assertive, collaborative, questioning, etc. — across their turns. Useful for coaching and performance reviews.
Your transcription balance is zero. Free tier users: your 30 min resets on the 1st of the month. PAYG users: click "Top Up" in the nav bar to add credit.
This usually happens with very large files over slow connections. Try: (1) compressing the audio to MP3 at 128 kbps before uploading, (2) splitting long recordings into 30-min chunks, or (3) uploading to Storage first and using the URL path.
Diarisation can struggle when speakers have similar voices, frequently interrupt each other, or when background noise is present. Use the speaker naming panel to manually correct labels after transcription.
Check that your browser has not blocked scripts from payfast.co.za. Disable any ad-blockers for this site and try again. If the problem persists, try a different browser.
This occasionally occurs if the transcript is very long (> 100 000 words). Try exporting to clipboard or SRT first. If DOCX consistently fails, log a support ticket.
Ensure third-party cookies are allowed in your browser for this site, as Google OAuth requires them. Safari users may need to disable "Prevent cross-site tracking" in settings.
Can't find the answer here? We're here to help.
Report a bug, billing issue, or ask a question. We review all tickets and respond within 1 business day.
For general enquiries or partnership discussions, email us at info@techsight.co.za and we'll get back to you.
Note: LinguistPro is a product of Tech Sight (Pty) Ltd, registered in South Africa. Support is provided in English and Afrikaans.