What audio extraction really does, why 192 kbps is the sensible default, when converting is a bad idea, and how to keep a recording sounding the way it was recorded.
Turning a video into an MP3 sounds like it should be simple, and mostly it is — but there is one decision in the middle that determines whether the result sounds fine or noticeably worse than the original. It is worth understanding before you convert a hundred files.
Extraction versus conversion
A video file is a container holding a video stream and an audio stream side by side. Pulling the audio out of the container is extraction, and it is lossless: the audio data is copied untouched. Converting that audio into MP3 is a separate step, and it is not lossless — MP3 is a lossy format, so re-encoding discards information.
Online video almost never carries MP3 audio natively. YouTube serves Opus or AAC, so producing an MP3 means decoding the original and re-encoding it. That is a second lossy pass on already-lossy audio, which is called generation loss.
So how bad is generation loss?
At a sensible bitrate, not bad at all — which is why we encode at 192 kbps rather than something lower. At 128 kbps the second pass starts to be audible on music, particularly on cymbals and reverb tails. At 192 kbps the vast majority of listeners cannot pick the converted file from the source on ordinary equipment, and speech is effectively untouched.
| Bitrate | Size per hour | Verdict |
|---|---|---|
| 96 kbps | ~43 MB | Speech only, and only when space is tight |
| 128 kbps | ~58 MB | Fine for podcasts, audibly thin on music |
| 192 kbps | ~86 MB | Our default — transparent for most listeners |
| 320 kbps | ~144 MB | Larger files, no real gain from a lossy source |
When you should not convert
- When you already have the original recording. Always work from the master, never from a re-encoded copy.
- When you plan to edit the audio heavily. Every save re-encodes and compounds the loss; edit from the least compressed source you have.
- When you do not have the right to the audio. A song is protected whether it arrives as video or as MP3, and stripping the picture changes nothing about who owns the recording.
Where audio extraction is genuinely useful
Lectures and conference talks you want to listen to while commuting. Podcasts that were published to video platforms and nowhere else. Interviews you need to transcribe. Your own recordings, where you want the sound without the picture. In each case the value comes from changing how you consume something you already have the right to — which is the honest use of this kind of tool.
About this article. Written and reviewed in-house by the All-in-One Video Download team at Parv Gupta. It is not sponsored, contains no paid links, and no advertiser had any input into it. Spotted an error? Tell us at support@allinonevideodownload.com and we will correct it.
General information only — not legal advice. See our Disclaimer.