Use open-source OGG Vorbis instead of MP3

1

Here’s why

Export a video with whatever mp3 bitrate you usually use.

Then, try using variable and also average bit rate options.

Here are some bit rate equivalents for OGG since Audacityvuses the quality 0-10 scheme instead of raw bitrate information: My data comes from a post on the Audacity forums.

Q0. If you use spotify for music, turn on data saver–don’t tell anyone else listening with you, and see if anyone notices. Then switch to Pandora free which uses 64 bit mp3. OGG is superior.

Q0 is average 64 kbit

Q3 is somewhere around 110 kbit average.

Q5 is the default, at 160 average.

No idea what the rest are, but Q10 is near 400 average, about half that of most 800-900 kbit/s .flac files.

So what I would like you all to do is do tests comparing usual mp3 audio settings, and compare that with:

Quality 0, 3, and 5 OGG files.

Since this is just speech, it will be interesting to know of and when you start to hear a difference. I complained about the sound quality to Noah on his show, when he made an S sound, it sounded wavery, or distorted. But only on that sound. He didn’t reply with an email, but after I sent it, there was no longer that issue in the audio in any episode after that, so I know he made some change.

There is no noticable audio issues on your shows, Destination Linux, Linux Out Loud, or any others. I will mention that on TWIL, there is an effect (I am guessing some compressor or limiter, but I don’t know what O’m talking about) when Michael first starts a segment, it sounds garbled, bur doesn’t last but a few words. That’s not something I would consider an issue.

I’m just thinking what could be: even higher quality audio, with a superior audio format with similar or slightly higher bitrate. The really annoying thing is that Apple devices don’t natively support OGG. I can’t understand that, and it’s another reason to never use iOS. AAC isn’t bad, no, it’s a very good format, much better than mp3, but to actively prevent a user from using an open audio format is REALLY odd.

Would you consider having a separate OGG archive for those that want it?

You spend a lot of time harping on HDMI being proprietary, why not stop supporting MP3? iOS users are still left out, as AAC is the only lossy format supported by iOS that isn’t mp3. There is a Linux podcast that already offers OGG versions of the show, my memory is hazy though, maybe it is LWDW, aka Linux game cast, but that might not be it.

I’m quite amazed that you got booth number 420, and I sincerely thank you for throwing in all the pot references, not just dope, there were a few more I can’t recall that seem like you intentionally threw it in. I really appreciate that.

2

Going Linux has an OGG feed as an option. Maybe that’s the one you’re thinking about.

It’s very easy to produce your own OGG versions with ffmpeg.

ffmpeg -i Linux_Out_Loud_Episode_181_Everything_Matt_Hates.mp3 Linux_Out_Loud_Episode_181_Everything_Matt_Hates.ogg
3

This will never acheive a higher quality. It is not strictly havong it in PGG format, rather using OGG or I now recently learned about opus, because at bitrates as low as 32 kilobyte/s rates, music encoded into mono output, is excellent quality. This mist be done at recording time Re-encoding a shitty proprietary format into any other format will never improve audio quality

MP3 cannot do that. I even tested mp2, as some people say it does a better job, basically the same at very low bitrates.

Speaking of Linux out Loud, the latest episode (I’ll redownload several other ones to vheck) seems to be lower than usual. This is odd because I can tell just by lookong at the filesize, that the bitrate is 160 kbps. How do I know? Because 128 constant bitrate basically gives you song length in megabytes. Of a song is 4:20 then you’ll get a file VERY close to 4.20 and can tell how long a song is just by looking at filesize.

I encourage you to download some .FLAC files online, or, 320 kbps mp3 from artists who release their music for free, lile Pretty Lights.

Then use

opusenc --bitrate 32 --cvbr --downmix-mono '/path/to/your/FLACfiles/song.flac' '%f.opus'

You’ll be quite amazed how good a 32 kbps audio file can sound. For stereo, it just needs at minimum, double the bitrate to be comparable, but around 55 kbps is not much to complain about, neither is 64. About 78 to 80 seems to be the sweet spot for excellent quality, and quite small files.

After using openenc, open Audacity with a .flac file, and encpde (export) as MP3 with 80kbps, 96 kbps, and variable 90-135).

1 Like
4

Yes. This is very interesting. The flac that I used is around 95 Megs with 5,000+ kb/s, the ogg is about 2.5 Megs but the opus is 950K with around 49 kb/s – Yet the opus sounds amazing.

Thanks for the tip.

5

EXACTLY!!

I apologize for my abrasove comments, but this is the difference in quality opus can make. I don’t see a benrfit to exporting in stereo for a podcast, unless the host wanted one voice on left, and one or two on right speaker.

So exporting to mono for a podcast and 32-48 kbit rate sounds WAY better than mp3 even at 96 or 128. Shocking isn’t it?

Pleaase, take that ridiculously high bitrate flac file (haven’t seen one above 900 kbps or so) and convert to mp3 now.

Forst try 64k average and constant. Awful, next tty 80 average and constant, then try 128 average and constant, and tell me the 128 kbit mp3 sounds better than thr 32 kb rate opus in mono.

I’ve surprising nevervtried exporting mp3 in mono, so I’ll give that a go, but I will not have high hopes for sound qualoty. Also the Linux out Loud episode is lower quality than usual, and there have also been a few episodes without one host that uses lower bitrate, where it sounded similar to this one, at 160 kbps. I know the team is very, very busy, and ot’s very uncommon for this to happen, sp it’s not at all a complaint, just an observation.

I think getting away from MP3 can solve this issue, at at even 96 KB?s, there was a 50+ minute epiaode at 40 MB, opus at this rate is plenty for stereo musoc, so speech in mono, wow, I bet that must sound great.

ALSO, opus specifically has a setting for speech, instead of music:

opusenc --speech --downmix-mono --botrate 96 or even 48 'input_file/path' '%f.opus'

So, I hope this suggestion goes somewhere but how will opus be accessible on iOS devices? It’s so annoying.

6

No worries. I always tried to read online comments in their best light and try not to default to ill intentions. Life is less stressful that way.

Using Mono was a practice I established way back in the early days of podcasting – I’m talking before there were smart phones or broadband. Making podcasts in Mono never affected the quality but reduced the audio size by half. Back when everyone was on dial-up, it was an important consideration. Creators now tend to believe all of their listeners have the same (24/7 connection to broadband) or similar setup as they do. It’s a common mistake.

This is probably the killer. IIRC ogg/opus has no method of DRM and for that reason, I would be surprised if Apple ever supports it. Not that DRM is used or needed here, but it seems to me that Apple would never open that door.