Non-destructive MP3 editing

Audio Plugin Hosts and other audio software applications discussion
Post Reply New Topic
RELATED
PRODUCTS

Post

A few years ago, I stopped recording my source material in wav. I use flac instead, and it works fine for me. However, my field recorder only records in wav, or MP3 — though I can record at 320kbps.

I like having field recordings in MP3 because I can store and organize them as easily as my other MP3s. My only hesitation is that I'll lose sound quality if I ever try to use them in a podcast, a song, or some other purpose.

So I have two questions:
  • Is there a recommended program for editing MP3s in a non-destructive way? I use Linux, and Audacity is destructive. If I open an mp3 to trim it, the program will decode it, then re-encode it. FFMPEG is a linux library that does non-destructive MP3 editing, but I have yet to find a good GUI for it. Furthermore, I'd like to fade the ends of my recordings, and I don't know if there's any way to do this withoug re-encoding.
  • How much damage is actually done if I'm importing MP3s at 320k, tweaking them (normalizing, compressing, editing), then re-encoding as a 320k MP3? I suspect audio compression would expose audible data compression worst of all, but again, I don't know how perceptible this is at 320k.

Post

This is a good little app without decoding: http://mp3splt.sourceforge.net/mp3splt_page/home.php
Image
Intel® Core™ i9-9900K•Cubase 11•Presonus Eris E8 XT•Focusrite Scarlett 18i20 & Octopre•NI Kontrol S61 MK2•Stein­berg CC121•Synthesizers: Arturia Casio Korg Roland Yamaha

Post


Post

Thanks mladi, mp3splt was the best thing I could find for native linux.

No_Use, mp3DirectCut looks like the best solution. Even though it's not native linux, it looks like the author develops with Wine in mind. And the fact that you can normalize and fade without re-encoding makes this a great option. Thanks a lot.

Maybe someone can speak to the other question: is it a bad idea to work in 320 MP3s? I would never record for music this way, because of how I process that kind of audio (for instance, taking a line-in electric guitar recording, then running it through a stompbox emulator, then adding compression and tape saturation would start to bring out those compression artifiacts). But if you're recording found sounds, and maybe you want to make a sound collage or something, and you're only processing and re-encoding once, is it still a mistake to work with MP3s?

Post

I would recommend recording in a lossless higher quality format. Not sure what “field” it is that you are recording in :D but 24bit/44.1 or 48k is going to sound better especially after editing. When i record birds or bugs i go to 96k, and most people can hear the difference (on a good system in my testing lab.) :)

You can always batch file convert them to mp3 later for convenience.
gadgets an gizmos..make noise https://soundcloud.com/crystalawareness Restocked: 3/24
old stuff http://ww.dancingbearaudioresearch.com/
if this post is edited -it was for punctuation, grammar, or to make it coherent (or make me seem coherent).

Post

One should always aim to work at the best quality if they can, only converting down as the last step for the format needed. But if it doesn't bother you, and it's not practical, and it mostly sounds fine, then that's your call. It's not ideal, but don't let it stop you from working or releasing projects. Sometimes convenience is inspiring and perfectionism is debilitating. It seems like sometimes people are afraid of things they can not hear, and that some golden eared police will expose them, and toss them in audiophile jail.

Just general thoughts, as I have no knowledge of non-destructive mp3 editing to offer.

Post

Thanks for the replies. CrystalWizard, I usually record at 24bit/48k. Maybe I should record at 96k for things like bugs. I've read that if you record at 96, then downsample to 48, the quality is still better — more gets picked up on the recording.

I think if I had a field recording, and there was something interesting in there, but barely perceptible — like a distant conversation — the best chance of isolating and enhancing that would be with a lossless file.

Considering .flac will bring the file size down 40%—50%, and that you can tag, organize and play them the same as MP3s, then it sounds like the best move is to record 24bit / 96k .wav files, then batch convert them to flac at the same sampling and bitrate.

Post

Keep in mind also that I suspect .mp3 compression will be filtering out sounds above 15kHz or so

Post

publicradio wrote: Mon Jan 20, 2020 10:51 pm
I think if I had a field recording, and there was something interesting in there, but barely perceptible — like a distant conversation — the best chance of isolating and enhancing that would be with a lossless file.

Considering .flac will bring the file size down 40%—50%, and that you can tag, organize and play them the same as MP3s, then it sounds like the best move is to record 24bit / 96k .wav files, then batch convert them to flac at the same sampling and bitrate.
:tu:
gadgets an gizmos..make noise https://soundcloud.com/crystalawareness Restocked: 3/24
old stuff http://ww.dancingbearaudioresearch.com/
if this post is edited -it was for punctuation, grammar, or to make it coherent (or make me seem coherent).

Post

zacha wrote: Mon Jan 20, 2020 11:32 pm Keep in mind also that I suspect .mp3 compression will be filtering out sounds above 15kHz or so
I've done some tests in the past, but this is only true for lower bitrates. Lower than 128kbps iirc. 320kbps should go all the way up to 22 kHz. But it may depend on the encoder.

What I would do, is keep the original .mp3 as archive and convert to WAV as first step whenever you're actually using it.
We are the KVR collective. Resistance is futile. You will be assimilated. Image
My MusicCalc is served over https!!

Post

Uh, no, i would suggest opposite. Record in wav, edit in wav, then degrade into mp3. This way you have the quality source, you can edit the quality, then just mp3 what you need to be small files. With the cost of digital storage these days there is no reason to skimp on quality.
gadgets an gizmos..make noise https://soundcloud.com/crystalawareness Restocked: 3/24
old stuff http://ww.dancingbearaudioresearch.com/
if this post is edited -it was for punctuation, grammar, or to make it coherent (or make me seem coherent).

Post

Check out https://rogueamoeba.com/fission/ Mac Only...

Post

publicradio wrote: Mon Jan 20, 2020 10:51 pm Thanks for the replies. CrystalWizard, I usually record at 24bit/48k. Maybe I should record at 96k for things like bugs. I've read that if you record at 96, then downsample to 48, the quality is still better — more gets picked up on the recording.

I think if I had a field recording, and there was something interesting in there, but barely perceptible — like a distant conversation — the best chance of isolating and enhancing that would be with a lossless file.

Considering .flac will bring the file size down 40%—50%, and that you can tag, organize and play them the same as MP3s, then it sounds like the best move is to record 24bit / 96k .wav files, then batch convert them to flac at the same sampling and bitrate.
1) Technically if you downsample 96 KHz to 48 KHz you will end up with the same 23.X KHz hard limit... So the ending result will not have more information than as if you recorded at 48 KHz.

2) MP3 at 320kbps doesn't truncate everything below a certain threshold, but a lossless file would generally be better for that purpose yes.

3) Yes. I sometimes do that with my logic projects - convert the whole audiobin to lossless CAF for archiving.
Image

Post Reply

Return to “Hosts & Applications (Sequencers, DAWs, Audio Editors, etc.)”