What you get from an MP3 file
MP3 is still the format most recordings end up in. Podcast episodes, voice recorder apps, exported phone calls, lecture capture systems and old interview archives all produce it, which is why "mp3 to text" is usually the first thing people search for when they need a transcript. Transcrify takes the file as it is. There is no need to convert it to WAV first or to trim it in an editor.
The result is a transcript split into paragraphs, with a timestamp at the start of each segment and a label for every speaker the model could tell apart. The language is detected automatically from the first seconds of audio, and more than fifty languages are supported, so a bilingual interview or a podcast recorded in Spanish needs no special settings.
Compression matters less than people expect. Speech survives MP3 encoding well, and a mono file at 64 kbps transcribes about as accurately as a studio master. What does hurt accuracy is the recording itself: a phone lying on a table three metres from the speaker, heavy background music or two people talking over each other. If you can choose, record close to the mouth and keep the music for the final mix.