By Ravi Kumar
In this Modlingua Live session, Ravi Kumar, founder of the Modlingua Group, explains a rule of thumb for transcription work and what changes when you transcribe from one language into another. This is part two of the "golden rules" lesson. The video is titled "I Spent 30 Days Mastering Transcription Fundamentals and Got SHOCKING Results" on YouTube.
The six-times rule
The speaker's rule, from his own experience and that of other transcribers, is that one minute of audio takes at least six minutes of work. Half an hour of audio could therefore need about three hours. He suggests multiplying the audio length by six to estimate the time to quote, and then setting a price.
Pricing, as the speaker sees it
He says rates in India are often lower, around one to two dollars per audio minute, while international rates start around three to five dollars. He adds that your price depends on your skills, availability, the market, the client's budget and how much time you have. These figures are his remarks, not a current rate card, so check market rates before quoting.
Word-for-word transcription
Transcription means writing down what is said, word for word, while keeping the sense. You do not add your own creativity, but you do need presence of mind so the message comes across. Transcribing within one language, for example Hindi to Hindi or English to English, is routine work. Transcribing from one language into another, such as Hindi into Tamil, Telugu, Punjabi, Spanish, French or German, brings in translation skill and a heavier mental load, which is why the speaker says it deserves higher pay.
Why the skill opens doors
Video consumption is rising worldwide, and a video in one language is likely to be needed in many others. The speaker shows how a transcript can be uploaded to YouTube, which then times it against the audio, so you do not need to add timing tags by hand.
An example with several languages
In his example, an audio file was transcribed from Gujarati into English and then translated into Spanish. Poor audio is marked, for instance "female, poor voice quality", and speakers are identified with timings so readers know what happens when. He stresses that the first transcript must be clear and accurate, because loss of meaning builds up with each language step. Loss cannot be avoided completely, but it can be kept small.
Questions to take away
- How many hours would a 20-minute audio file take you at six times its length?
- What extra checks would you add when transcribing between two languages?
- Which markers would you use for unclear audio or several speakers?
Related reading
For a practical guide, read Transcription and subtitling: a practical guide for translators.
