Whether you're pulling a podcast recording out of a screen-captured video, saving a lecture's audio to listen to offline, or grabbing a song from a music video, this strips the video track and keeps just the audio, encoded as a standard MP3. The whole conversion runs through an in-browser build of FFmpeg, so even long recordings process without the video ever leaving your device. For a dedicated audio-extraction workflow with more format options, see DuckConvert's Extract Audio tool.
You cannot get back more than the video had
The audio sitting inside an MP4 has already been compressed, almost always as AAC, and whatever detail that compression discarded is gone before this conversion starts. Producing an MP3 means running that audio back through a decoder and then out through a different encoder entirely, so the result can come close to the source but can never better it.
The practical consequence is that a high encoder setting is worth using and a very high one is wasted. Past a certain point you are spending bytes faithfully reproducing compression artefacts that were already in the video.
How much smaller the result will be
Almost all of a video file is the picture. Strip it away and what remains is usually somewhere between a twentieth and a tenth of the original size, which is why a two-gigabyte recording of a conference talk becomes a podcast-sized file you can keep on a phone. At a typical variable bitrate an hour of audio lands around 80 to 90 megabytes.
The video's resolution has no bearing on any of this. A 4K recording and a 720p recording of the same event produce audio files of the same size, because the audio track was encoded independently of the picture and is all that survives.