Upload an interview, meeting or lecture where people move between two languages. The language is detected automatically, with no setting to choose, and you get text with speaker labels and word-level timestamps. Files up to 10 hours or 5 GB; files over 500 MB continue on the upload page.
How it works
Mixed-language audio always deserves a review pass. The steps build it in.
Drop the file here (up to 500 MB; larger files continue on the upload page) or paste a public YouTube link. No language to select.
Read the places where speakers switch languages and listen at the timestamp wherever a word looks wrong.
TXT, DOCX, SRT, VTT, JSON or MD. The text stays in the languages that were spoken.
What to expect
Real conversations switch languages mid-sentence. Here is what happens with those recordings.
The language is detected from the audio for each recording. Where a recording mixes languages, the aim is to write each part in the language actually spoken.
Hinglish comes out with the Hindi in Devanagari and the English in Latin script, matching how people talked. Heavily mixed speech needs a review pass.
Taglish, the everyday switching between Tagalog and English, is handled. As with any mixed recording, check the passages that matter.
Results on mixed audio depend on how often speakers switch, the language pair and the audio quality. Check the transcript against the recording before relying on it.
Practical advice
A little preparation makes mixed-language transcripts noticeably cleaner.
If a recording has separate sections in each language, such as a talk followed by a question session in another language, split them into separate files before uploading.
If a recording opens with a long stretch in one language but is mostly in the other, uploading that opening as its own file can help.
Background noise and people talking over each other hurt mixed-language audio more than single-language audio. Use a close microphone.
Whipscribe does not translate. The transcript is in the languages spoken, so plan any translation as a separate step.
Pricing
A pack from $4 opens the full transcript, and anyone can preview instantly with no signup. After that, one-time packs cost $8 for 1,000 minutes, $12 for 2,000 minutes or $24 for 5,000 minutes. Credits never expire and there is no subscription: about $0.29 to $0.48 per audio hour, depending on the pack.
| Recording | Length | $8 pack | $12 pack | $24 pack |
|---|---|---|---|---|
| 30-minute interview | 30 min | $0.24 | $0.18 | about $0.14 |
| 1-hour meeting | 60 min | $0.48 | $0.36 | about $0.29 |
| 3-hour workshop | 180 min | $1.44 | $1.08 | about $0.86 |
Costs are the share of each pack the recording uses. Mixed-language recordings cost the same as any other. See pricing.
FAQ
Yes. The language is detected automatically, and where a recording mixes languages the aim is to write each part in the language spoken. Accuracy on mixed audio varies, so review the transcript.
No. There is no language setting; the language is detected from the recording automatically.
Yes. Hindi comes out in Devanagari and English in Latin script, matching how it was spoken. Heavily mixed speech needs a review pass.
Yes. Taglish is handled. Check the passages you plan to quote or publish.
No. Whipscribe does not translate. The transcript keeps the languages that were spoken.
It varies with how often speakers switch, the language pair and the audio quality. Check the transcript against the recording, especially names and quick switches mid-sentence.
Where a recording has separate sections in each language, one language per file usually gives a cleaner transcript. For conversations that switch constantly, upload the whole file and review.
A pack from $4 opens the full transcript. After that, a 1-hour recording costs $0.48 on the $8 pack, $0.36 on the $12 pack, or about $0.29 on the $24 pack. Credits never expire.
Instant preview, no signup, no card. A free account keeps your transcripts in one searchable library.