Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: MP3 Conversion analyzer (Read 2892 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

MP3 Conversion analyzer

Hey folks,

I convert 44kHz 16bit Mono WAV files to 96Kbps MP3s. I'm looking for a utility that would analyze an mp3 and compare it to its originating WAV. A graphical app would be great if it could show the difference between the original and the converted MP3's waveform.

Thanks in advance
Zooya

MP3 Conversion analyzer

Reply #1
Hi Zooya,

visual comparison of waveforms is not the important point when it comes to evaluate the quality of lossy compression. The trick behind lossy data reduction is to deform the original signal, but to keep the deformation at places where the are not too obvious.
The only valid way are listening tests

(note that the above is a brazen simplification of the real thing!)

MP3 Conversion analyzer

Reply #2
To reiterate Sunhillow's comments above, looking at waveforms will tell you very little. As is often said by people around here, you listen with your ears, not with your eyes. That may be more true than you realise, so I've listed a few possible alternatives that you may find helpful if you want to make the most of the MP3 format at this kind of bitrate.

You might like to try the LAME MP3 encoder in VBR (variable bitrate) at -V7 with no additional switches. This would generate a Joint Stereo encoding at around 100Kbps that may sound better to you than CBR96 in mono. This also eliminates the potential problem of a forced mono encoding being accidentally played back at double-speed by some players. Joint Stereo encodes truly mono content very efficiently with the advantage of being able to preserve any "accidental" stereo with high efficiency also if it occurs in the original content.

I'm always interested in how others get on in these circumstances as I encode original audio CDs of stereo radio shows at -V6 or -V7 (depending upon content) and have always been very pleasantly surprised by the result in what seems to be an infeasibly small bitrate for MP3. VBR should nearly always give the best possible ratio of quality versus bitrate, so it makes sense to run a model that targets quality rather than a specific bitrate when we're already a little bitrate-starved, IMO. Maybe the quality is even better with predominantly mono content.

If the 32KHz resample that happens at -V7 bothers you then you might like to go up a step to -V6 where no resampling takes place. This would increase the typical bitrate to around 115Kbps as well as raising the imposed low-pass filter frequency slightly, and this may sound noticeably better to you personally than -V7 or CBR96 for a relatively small increase in average file size. I have many encodings here at -V6 that make occasional usage of 256Kbps as and when deemed necessary, whilst they maintain the ability to drop all the way down to 32Kbps for sections of perfect digital silence, yet overall, they average around 115Kbps.

Another option worth considering at this bitrate range is ABR (average bitrate) encoding. It has a little of the bitrate flexibility of VBR coupled with an almost guaranteed average bitrate meeting the intended target bitrate, so ABR96 or ABR112 might interest you if target bitrate is more important to you than outright target quality. ABR also encodes in Joint Stereo by default.

You may even prefer CBR96 in its default of Joint Stereo in comparative blind listening tests.

When working at bitrates where total perceptual transparency is statistically unlikely at best, try to think of the choice of encoding strategy as being similar to selecting a tape formulation to make the most of a high-quality cassette deck. Just pick whichever sounds nicer to you, and bear in mind that different source material may benefit from different strategies. But please, ignore the waveforms! Only you have your ears.

I'd highly recommend having a play. It's good fun, honest.

Cheers, Slipstreem. 

MP3 Conversion analyzer

Reply #3
Thanks for your answers guys. To reply Slipstreem: I'm afraid of using VBR or ABR at all, because at some media players they can cause malfunction especially when seeking in streams (rewind, forward) and displaying time. The only solution for me is CBR, 96kbps is good enough, but I need a really good encoder. I don't think lame would do the best job in default mode. I mean, I tried it and could hear the difference. I don't want to raise the bitrate, because it's still mono and 44kHz, but I get surprised when 64kbps (also mono) public radio podcasts sound so great while they are CBR.

I can use my ears, yeah, but there are some cases when I can't hear the difference any more, because the compressed material's sounding is too close to the originating one. It's like when you can't see JPEG artifacts on a picture just by looking at it, but if you manipulate (equalize, change hue&saturation) you begin to clearly see them. That's why I'd need some application which tells (or shows) me the difference.

 

MP3 Conversion analyzer

Reply #4
If you want to later edit the audio file, it is recommended that you keep it in lossless format.
Lossy formats remove/modify parts of a signal that are expected to be inaudible when listening, but without further editing.

An Example:
Among other things, lossy uses properties of masking, thus part of audio at lower frequency will mask some parts at higher frequencies. Because of this higher frequencies can be modified to allow more compression without artifact that could be heard. If you afterwards remove lower frequencies from the original and from the compressed signal, you'll definitely hear a difference in the remaining high frequencies.

On the other hand, with higher bitrate, more of the original signal is preserved and thus more editing is possible without noticing artifacts.
You might do editing before using lossy compression or increase bitrate until edited output of lossy if free of artifact that you can hear.