Audio Transcription with AI
Upload a recording and let our AI audio transcription turn it into a clean transcript as a PDF. Works in dozens of languages with automatic detection - and when a recording alternates between languages (a speaker and their interpreter, a bilingual class), each language is split into its own labelled track. 2 hours of transcription are free every month.
How it works
- Sign in, then upload a recording (MP3, M4A, WAV, OGG, Opus or FLAC).
- Leave the language on auto-detect - or pick one if the whole recording is in a single language.
- Press “Transcribe”. The AI transcribes the audio to text chunk by chunk, and every turn is labelled with its detected language.
- Pick a layout - continuous, two columns, or main language with notes - then download the transcript as a PDF or a text file.
Frequently asked questions
Which languages are supported?
Dozens - the speech model covers most widely spoken languages, and the language of each passage is detected automatically. If the whole recording is in one language, selecting it explicitly improves accuracy.
How are multiple speakers handled?
Recordings that alternate between languages - a speaker and their interpreter, a bilingual class - are split automatically: each language becomes its own labelled, color-coded track, in the on-screen preview and in the PDF. Speakers sharing one language are transcribed as a continuous flow.
How much does it cost?
Every account includes 2 free hours of transcription per month. Beyond that, two options: RectoPDF Pro at €6.99/month (or €69.99/year) with 30 hours included every month, or a one-off hour pack with no subscription - 10 hours for €4.99 (20 hours for €4.99 with Pro). Purchased hours never expire.
Is my audio uploaded anywhere?
Your audio is sent securely to our server only to be transcribed; it is not kept afterwards. The transcript and the PDF are then assembled in your browser. You need to be signed in to use it.
Does it handle religious recordings?
Yes - recited Quran verses and quoted hadiths are recognized, matched against the authentic text and shown with their exact wording and a reference (Surah:Ayah, or collection and number), and common invocations are typeset in their proper form. This runs automatically on Arabic content. Quran text: Tanzil project; hadith text: sunnah.com.
Is this an AI audio transcription tool?
Yes. Speech recognition runs on a large multilingual AI model, and an optional AI pass corrects misheard words, spelling and punctuation. You get accurate audio-to-text transcription without installing anything.