Skip to content

Urdu transcription

Urdu audio and video to text

Upload Urdu audio or video and get an editable transcript in Urdu script. Be realistic: Urdu is one of the weaker languages for Whisper-family models, because far less Urdu audio was available for training, and the small model that runs here will make many mistakes. Treat the result as a first draft and use short, clean, single-speaker clips.

Free preview: one file up to 15 minutes and 25 MB, 3 tries a day, no account needed.

Want to keep your transcripts? Create a free account

Urdu, Hindi and Hindustani

Everyday spoken Urdu and Hindi are largely one language, Hindustani. They differ mainly in script and in formal vocabulary, with Persian and Arabic words in Urdu and Sanskrit-derived words in Hindi. That overlap is why a transcript can drift: the model may spell a word in a Hindi-like way, and if the language were not set it could treat Urdu as Hindi. This page presets Urdu, but check word choices. Pakistani, Hyderabadi (Dakhini) and diaspora Urdu differ, and accents from Punjabi, Sindhi, Pashto or Saraiki speakers add more difficulty.

Script, right-to-left and Nastaliq

Urdu is written right to left in an extended Arabic-based script: اردو. The output is plain Unicode text. Nastaliq is the calligraphic style traditionally used for Urdu, but most apps show it in a plainer Naskh-style font unless a Nastaliq font such as Noto Nastaliq Urdu is installed. The words are the same; only the look changes. Only Urdu script is produced, not Roman Urdu typed in Latin letters. Punctuation (۔ and ؟) and digits may be inconsistent, and English words inside Urdu sentences can shift the display order.

What speakers of Urdu use it for

Students record لیکچر from universities, madrasas and online classes. Journalists and researchers transcribe interviews. Families keep WhatsApp voice notes as text. Creators draft captions for dramas, vlogs and short videos. Others transcribe religious talks and sermons for study, where every word must be checked before it is quoted. In all cases the result is a draft to proof-read, not finished text.

Which model works best for Urdu

Urdu is among the languages Whisper-family models have seen the least speech in, and the small model is often only a rough guide. Use short, clean, single-speaker clips and correct the draft against the audio. The larger models help when you self-host Scriptorium, but even they make many errors in Urdu.

On tryscriptorium.com the no-account preview and regular accounts run the small model (the larger medium and large-v3 models run when you self-host Scriptorium). In an account, pick Small in the model menu rather than Auto: Auto prefers the fastest, least accurate model for long files or a busy queue.

The small model gives Urdu the least help of any language here, so recording quality matters more than usual: record close to the microphone, one voice at a time, with little music, in short files. If the transcript matters, self-hosting with large-v3 is the stronger route, but it still needs a proof-read.

Subtitles and captions in Urdu

Urdu subtitles run right to left, so test the SRT or VTT in your player before you publish. VLC and YouTube read UTF-8 and show Urdu script, but some editors need a right-to-left setting or an Urdu font. Keep lines short, since Nastaliq-style text is tall. Proof-read every line: a wrong word in a caption is more visible than in a text file. Subtitles stay Urdu.

The uploader is set to Urdu

The upload box on this page transcribes as Urdu, so the recogniser does not have to guess the language from the first seconds. If the recording is in another language or switches languages, use Audio to text, which detects the language automatically. Automatic transcription is a draft and no accuracy figure is promised: check names, numbers and anything you will rely on.

How it works

  1. Add a recording

    Drop in a video or audio file, choose one from your device, or record from the microphone. Nothing to install.

  2. It transcribes

    Speech becomes timestamped text, with speaker labels when more than one person talks.

  3. Read, fix, export

    Click any line to correct it, then download TXT, subtitles, Markdown, JSON or CSV.

Questions

Is the output Nastaliq or Naskh?
Neither is chosen by the transcript. You get plain Unicode Urdu, and the style depends on the font in the app that shows it. Install a Nastaliq font such as Noto Nastaliq Urdu if you want that look.
Does it write Roman Urdu or Hindi script?
No. Output is Urdu script only, never Roman Urdu. Urdu and Hindi speech sound very close, so check that spellings are the Urdu ones you expect. For Hindi, use the Hindi language option.
How good is Urdu on this site?
Weaker than the strong-tier languages. Expect a rough draft with errors in names, fast speech and English words, especially on noisy audio. Short clean clips help; self-hosting with large-v3 is the stronger option. See how accurate is AI transcription.
Which file types can I upload?
Video: MP4, MOV, MKV, WebM, AVI, M4V, WMV, FLV, MPEG, 3GP, TS. Audio: MP3, WAV, M4A, AAC, FLAC, OGG, Opus, WMA, AIFF, AMR, CAF. Anything else the server can decode as audio or video generally works too; there is no need to convert first.
Is it free?
Yes — one file up to 15 minutes and 25 MB with no account, up to 3 tries a day. A free account keeps your transcripts in a library and lifts the guest limits; the current limits and paid plans are on the pricing page.

Want your transcripts saved and longer files? Create a free account · See pricing · All use cases

Try it with your file

No account needed for a first transcript.