Karaoke Forums

Hi everybody!

Last night I subbed my first song with a bitrate of 128kbps (The Heart of Rock and Roll) and as you'd expect it seems to be a better quality then the 56kbps I used to sub at. But I read that while 128kbps is sort of the average just-acceptable bitrate for a CD quality MP3 file I've found situations where I've listened to some subs that actually sounded better at 56kbps then even some of the rates above 128kbps. I actually heard one sub that I recall being at 320kpbs and I was disappointed because I couldn't even finish listening to it because of all the distortion.

Apparently, a lot of this quality has to do with the individual systems that we use when we listen to eachother's subs. Better systems do better with higher bitrates and lower quality sound systems do better with the lower bitrate. Recently I tried subbing a song at 96kbps which I figured would be better then the 56kbps but, instead, it sounded worse with a lot of "tinny" wobbly distortion coming from the treble side of the spectrum. Do any of you also notice this coming from some subs?

Now, I have also read that you may get a better quality 128kbps sound in mono then in stereo. I'm not quite sure why that is but I did experiment some and it seems to be true. Plus, is it really necessary to sub a stereo file when you haven't added any special stereo effects to the recording? In other words, if all you've got is your mix coming equally out of two speakers, is it really necessary to create and sub a stereo version of the song that will basically be twice the file size but accomplish the same thing as the mono version - two tracks coming equally together out of two speakers.

What do all of you think? Do you have a favorite bitrate that seems to play best on your system or one in particular that your system doesn't seem to handle well when listening to other people's subs? Maybe we can find some common bitrate factor that will make it more generally enjoyable listening for everyone and make our uploads a little smaller, which of course translates into quicker downloads and playing as well.

Looking forward to your comments - in fact, you might even say I'm just "biting at the bit-rate" to hear your opinions on this subject!  LOL

Hugs to all,

I do not sub songs here at any particular bitrate cos my limit is the file size of 3mb. Thus a 3 min song will have a higher bitrate as compare to a 5min song. I would if possible go for 192kbps which is near cd quality.

To get to the highest possible bitrate I go for joint stereo or mono. just a note; if you compress a wav file to at say, 128kbps in stereo and end up with a 4mb file size, converting the same wav file at the same bitrate in mono will not get the file size to 2mb but somewhere around 3+/ -mb.; it never is half.

_________________
I can neither confirm nor deny ever having or knowing anything about nothing.... mrscott

In my opinion, any files compressed using a commercial encoder will not sound good, I'm only using a Lame encoder with the latest dll or wma.  Mp3's compressed at 128k with Lame will sound better than 320k with a standard encoder, question of using the proper tools here!

Good information! Thanks, BS!  :yes:

Hi Micky,

You've definitely nailed the problem as to why two different people can be submitting songs at 128k and yet one sub sounds tinny and the other sounds okay - the encoder is really the culprit. I also use LAME by the way. I just found an EXCELLENT article on MP3 and how it works and doesn't work. I'm going to include the link to the page below but here are some interesting excerpts from the page that explains alot of the problem:

Many listeners accept the MP3 bitrate of 128 kilobits per second (kbit/s) as near enough to compact disc quality for them. This provides a compression ratio of approximately 11:1, although listening tests show that with a bit of practice, many listeners can reliably distinguish 128 kbit/s MP3s from CD originals. To some listeners, 128 kbit/s is unacceptably low quality. Even though differences may be perceptible, this is acceptable for some listeners in some listening environments, such as a noisy car or train.

A few possible encoders:

LAME first created by Mike Cheng in early 1998, it is a (by contrast to others) fully LGPL'd MP3 encoder, with excellent speed and quality, rivaling even MP3's technological successors.
Fraunhofer Gesellschaft: Some encoders are good, some have bugs.

The quality of MP3 files depend on the quality of the encoder and the difficulty of the signal which must be encoded.

Good encoders produce acceptable quality at 128 to 160 kbit/s and near-transparency at 160 to 192 kbit/s.

Low quality encoders may never reach transparency, not even at 320 kbit/s.
So it is pointless to speak of 128 kbit/s or 192 kbit/s quality, except in the context of a particular encoder or of the best available coders. A 128 kbit/s MP3 produced by a good encoder might sound better than a 192 kbit/s MP3 file produced by a bad encoder.

Additionally, it is important to note that this is subjective. A given bitrate suffices for some listeners but not for others. The numbers given above are rough guidelines that work for many people, but in the field of lossy audio compression, the only true measure of the quality of a compression process is to listen to the results.

Here's the link to the MP3 webpage: http://en.wikipedia.org/wiki/Mp3

The above Wikipedia article cleared up a TON of questions I had about MP3 and I'm reasonably sure that if I bump the bitrate on my subs up to 160 kbit/s and almost definitely to 192 kbit/s that any "tinniness" that some people may be hearing will disappear. The only problem of course is how the file size is going to increase and how that increase is going to cause problems within the constraints of SS's file restrictions at their various subscription levels. I just upgraded my subscription so I'll be able to do up to 5MB files pretty soon.

I'm almost tempted to go back to the 56kbit/s default Windows PCM format because I never received even a single complaint about tinniness and, in fact, even Percy who knows a lot more about this bitrate stuff then me said he was amazed at the sound quality I was getting at 56 kbit/s. But I'll try the higher k/bits first and see how it goes on my next sub.

Anyway, I strongly encourage others to visit the above link for more info on MP3. By understanding how it works and what our particular encoders best bitrate compression is, it can only make our subs more consistently enjoyable for everyone.

Good luck!

Here is a link that may interest some people.

http://arstechnica.com/wankerdesk/1q00/mp3/mp3-3.html[url]

This article; may be a bit techie for some; but a must-read if you are doing a lot of mp3 convertion. It compare diffrent encoders ie LAME, xing, FhG, and Blade. This is one of the best test comparision I have come across.
[/url]

_________________
I can neither confirm nor deny ever having or knowing anything about nothing.... mrscott

There are lots of 'tricks' to getting a good mp3.  Using a good encoder is first and foremost, but if you can use VBR (variable bit rate) you'll get better quality in a smaller file than using CBR (constant bit rate) at the same bit rate.

Since I'm limited by size on my subs here I tend to sub at the maximum bit rate I can use to get under the limit.  This means a shorter song gets recorded at a higher bitrate than a longer one.  I almost always sub in mono unless the song has specific stereo effects that are needed to get the full effect of the song.  Even then, most people don't set up the speakers properly to get good stereo seperation so it's a moot point.

I personally use LAME, but BladeEnc also works well.  BTW, I do my subs at CBR because Audacity doesn't support VBR encoding with lame.  I do encode my CD's using VBR and purchase most of my music in VBR at rates of 192kbps or greater.

Good points there on VBR and mono sub. But I too use Audacity and LAME and I can encode my sub using VBR. I have my LAME seperated and not link to Audacity. Mixdown to wav. name the file ,open LAME and get the file to encode.Thats how I do it. I do the same thing when using CRISTAL.

_________________
I can neither confirm nor deny ever having or knowing anything about nothing.... mrscott

Well, this thread is definitely working the "bugs" out of my understanding (or lack of) about MP3 compression. I just read a thread in an MP3 support forum that explained that you may create an MP3 file which sounds perfect on your system because your system is coding and decoding the MP3 with the same codec. So, for example, you can encode your MP3 with LAME (or one of the others), play the file and say "Okay, sounds great" and submit it to SS.  If your listener is using Windows Media Player (which is apparently more generally compatible with most codecs, including LAME) then they're going to hear the same quality that you heard and have no complaints either. BUT if they happen to have their browser defaulting to some other player that has an imcompatibility problem with your codec then they're going to hear distortion or they may not even hear the file at all if it's a severe incompatibility.

So this explains ALOT as to why members will sub a song which appears to have tons of distortion to the point where some of us may wonder "They're singing is so good but why would this person sub such a poor quality recording?" It's because it sounded just fine to them when being decoded by the same encoder they used to created the file.

From what I've read they only generalized solution is to create a file at a bitrate which gives a quality that is generally acceptable to most codecs on most systems and from what I've read that bitrate seems to be 160 and above but apparently there is no perceptible difference in quality to the average ear at bit rates above 256 kbit/s.

Also, it answers the question above why my 56kbit/s ACM files can play on most systems with less distortion then even a 128kbit/s file - because most systems apparently already have that ACM decoder installed and so the 56 kbit/s file is being properly decoded by the same software that encoding it in the first place.

So this appears to be generally a problem of software incompatibilities among various high bit codecs (those that go up to 320 for example) which have problems decoding MP3 files created at the lower bitrates. Perhaps the 160 k/bits is a general solution to the distortion problem but it still will leave most people with the file size problem where we need to keep our MP3s below a certain Megabyte size based on our various SS subscriptions.

Anyway, I for one am very pleased with how this thread has turned out so far. It's certainly helped narrow down the problem for me and reduced the level of frustration I sometimes feel when a listener says "the recording sounded tinny." (which is something I've said to others myself without understanding the root of the problem). So the MP3 wasn't "tinny" when it left my computer and it's encoder - it just "got tinny" when it was being read by the listener's encoder.

On a side note, Windows DirectX doesn't properly support VBR (although they might have fixed this recently).  It plays them just fine, but lacks the ability to accurately determine the length of the file (in minutes/seconds).

A problem?  Only if you're using a Windows based PC and relying on DirectX based programs to determine song length.  If you're using a non-windows Operating System or a product that uses some other means to determine song length then it doesn't affect you.

It's a little unusual to see a 'claimed' 2:30 song playing for over 3 minutes but you get to either ignore it or use a player that accurately reports the length.

BTW, I should have clarified, using the export as mp3 feature will only get you a CBR file.  To get VBR you would have to do as badsinger explained and use an external application.

I just use the default on my program (Goldwave) and I believe it is 128.  If the file size comes out too big, then I re-save it as 96 and that works.

_________________
~Suzanne Lanoue~

Hi all,

This thread has continued to help me understand more about "lossy compression" and I'm attempting to apply what I've learned here so far to improve the quality of my subs. I just subbed a song titled "Feeling Alright for SS Members". (Probably the perfect song to test bitrates on because it's a special paraody I wrote for us SS members)

The 160 bitrate created a file that was too large to upload (SS hasn't upgraded my account yet) so I used a program called Audiograbber (which is freeware) to create a 128 bitrate "Variable Bit Rate" file. It was really interesting to play the finished file and watch as the bitrate changed back and forth throughout the file from 128 to 160 to 192, etc and the final file was much smaller then the "Constant Bit Rate" 160 I had originally submitted. Also, I could not notice any discernable difference in quality between the 160 CBR file and the 128VBR file I ended up subbing.

When I subbed "The Heart of Rock and Roll" it was at 128 CBR so if you would like to do a comparison between 128 CBR and 128 VBR you may want to listen to both songs and see if you can pick up any difference. I think I'm going to stick with 128VBR for the time being but I'll wait and see if I get any major distortion reports from listeners.  But KM seems to be absolutely correct in pointing out the advantages of doing a VBR file save and if you can't do it with your present software, check out that Audiograbber program. It really seems to do the job and it's FREE!

All the best (including the best subs and the best ranks, of course) to you all!

I usually sub at 320/44 but my songs are shorter than most and my limit is 7 meg so for me a couple hundred bucks a year is worth if for the fun this site provides...hey I like that vbr deal, got to check that out....

Russ