They must be recorded literally each of the words of each person in the recording, respecting their language and dialect, and correctly including hesitations, false starts, interruptions, voice overlays, and so on.
If the speaker is expressed in Catalan, you must transcribe in Catalan. If you express yourself in English, you must transcribe in English. If it says "you", you must transcribe "you". If you speak of "vos", you must transcribe "vos". Do not write "you" when the speaker says "vos" or vice versa, and do not translate from one language to another.
If you find an English recording that has been transcribed into Spanish, either in some terms or the whole fragment, you should delete the translated parts and transcribe in the language of the recording. Transcribing from scratch and in the correct way, your contribution will be much higher and therefore you will receive a higher payment.