Solved MP3 "preview" plugin?

  • Thread starter Thread starter danny.guitar
  • Start date Start date
D

danny.guitar

Guest
I doubt anything like this exists, but maybe there's some kind of way to do it.

Since a lot of my music ends up as an MP3, it'd be nice to be able to add some kind of effect to the master buss that will let me hear it at "MP3 quality". Instead of having to render, listen, go back...render, listen, go back, etc.

I like to make separate mixes for MP3s that go on the internet.

Anyone know of a plugin for this? Or maybe some kind of EQ preset I can make that would be somewhat accurate?
 
Is there a way to take for example, white noise, save as WAV. Then convert it to an MP3...then compare the 2 and make an EQ filter based off that?

Would that work or no? And what software would be needed? I think Audacity can create white noise files but I don't know what to use for comparison/EQ.
 
MP3 does no EQ'ing and no noise either. It works on omission of frequencies.

Eventhough I'm not aware of such a preview plugin, it should be very possible to write one. All it needs is basically a vst which deploys the codecs available in the system. Or a vst which includes his own codec, which makes sense, since Window's mp3 codec is usually rather limited.
 
I know MP3s don't use EQ but I was hoping to mimic the result with an EQ filter. Or is this not possible?
 
if your stuff sounds good in the first place, it will stand up to mp3 compression just fine.

no need for a plug
 
giraffe said:
if your stuff sounds good in the first place, it will stand up to mp3 compression just fine.

no need for a plug
Click to expand...

Exactly what I was thinking.

Even if there was a big audible difference (which there shouldn't be), what are you going to do??? Make the wav sound worse so that the MP3 sounds better??? I figure the better your wav sounds, the better the MP3 will sound, no matter what.
 
danny.guitar said:
I doubt anything like this exists, but maybe there's some kind of way to do it.

Since a lot of my music ends up as an MP3, it'd be nice to be able to add some kind of effect to the master buss that will let me hear it at "MP3 quality". Instead of having to render, listen, go back...render, listen, go back, etc.

I like to make separate mixes for MP3s that go on the internet.

Anyone know of a plugin for this? Or maybe some kind of EQ preset I can make that would be somewhat accurate?
Click to expand...

I understand. There have been times I have converted a tune to mp3's and it doesnt translate right. Then I will convert a different tune and its fine. Yes it boils down to the MP3 codec you are using. I like Lame, but at times it does what I explained above. So I say it would be nice if say Lame or another codec (whatever its called) was in a plugin format and you could preview the file, tweak if need be then convert it.
 
It's completely different when you're recording pop/rock music, or a whole band, than just a single acoustic guitar track.

Completely different.
 
BushmasterM4 said:
I understand. There have been times I have converted a tune to mp3's and it doesnt translate right. Then I will convert a different tune and its fine.......
Click to expand...

I generally render my final .wav mixes to 24-bit and I've found when converting those to MP3 I sometimes get weird little dropouts. So now I downsample to 16 bit before encoding and that seems to work just fine.
 
It's not a simple EQ filter. The frequency rolloff is dependent upon both encoding rate and also level of high frequency content. You can see that if you watch an FFT while you playback a dynamic tune--you will see HF suddenly black as it reaches a low level.

It sounds like an interesting plug, I might give it a whack if I get some time in the next week or so . . .
 
I've also noticed that a lot of quick transients will introduce a lot of artifacts into the final MP3 file. For songs that have less (soft picking for example) or have gone through compression seem to come out a lot better.

Do you think that is the main reason for it sounding crappy? And maybe why pro (usually over-compressed) or pop/rock mixes sound better at MP3s? Less transients?

Some of my recordings come out sounding fine with almost no noticeable difference between the WAV and 128 KBPS MP3. Others, come out with the effects extremely noticeable. I am not doing anything significantly different between the recordings, other than the song I'm playing and how it's played. It's still usually just a single acoustic track. I think maybe the transients and high frequency content play a big role.

I've tried reading a Wiki article on MP3 compression but it doesn't seem to go into much detail.

If anyone could come up with a way to let me preview something as it would sound at say, 128KBPS MP3, I may actually pay for it. :D I'm really wanting to know more as a research kind of thing because I'm curious. :confused:
 
MadAudio said:
I generally render my final .wav mixes to 24-bit and I've found when converting those to MP3 I sometimes get weird little dropouts. So now I downsample to 16 bit before encoding and that seems to work just fine.
Click to expand...


I'll try the 16 bit and see how it does. Thanks.
 
I've found that encoding to mp3 can change the sound of the track a great deal, and not in a good way.

I think the idea of an mp3 preview plugin is brilliant, I'd love to see one! The plugin would need to simulate the various levels of compression that can be used with an mp3.
 
Let's not forget that not all MP3s are created equal. There are a couple of different codecs for MP3; Fraunhoffer and ISO not only do not sound the same, but even how they differ does not sound the same at differing sample rates.

If someone wanted to create such a plug, what they should do is they'd have to select a codec (hopefully both) and figure out how to actually execute them in real time. Trying to emulate the with filtering and such will wind up just like the scads of 1176 plugs which actually sound very little like the vast majority of actual physical 1176s out there.

But there's a big difference between the purpose of an 1176 plug and and the proposed MP3 plug is that an 1176 plug does not have to be accurate, it just has to sound pleasing in a similar fashon as the real thing. The MP3 plug, OTOH, by it's very purpose has to be accurate; the engineer needs to actually hear what it will sound like after conversion, not some kinda similar simulation of what it will sound like.

G.
 
I don't think it'd be too hard to load up the system codecs or whatever the user selects (LAME, Fraunhoffer, etc) then encode the data in real time. MP3 encoding/decoding is pretty light on the CPU and pretty fast.

If I was more familiar with C++/Steinberg SDK I'd make one, but I'm not. :(
 
Well there are a few barriers to exact replication of the various encoders. First of all, they aren't publicly released to my knowledge. Thus I would probably proceed with the sample encoding routines published in the standard, and compare that to results from commercial encoders, and perhaps add some controls to enable the user to alter the results to fit different encoders.

Also, mp3 encoding is essentially a look-ahead process: data is processed in frames, based on an analysis of the audio data in that frame. That is the cause of the phenomena I described earlier.

Whereas in a VST environment, a look-ahead process (that is already potentially computationally intense) would result in unacceptable latency. If you have to wait for a render, why bother with a VST? Just encode the file.

Also, mp3 encoding also involves a lossless compression after the frequency content is dynamically chopped. There is no point in modeling that function in a VST as the purpose is to monitor the output, not transmit the data over a limited-bandwidth network.

So I envision a tool that emulates the frequency compression behavior of mp3 encoding, with controls to select bitrate and quality settings with typical results. Is that worthless? Well, you get what you pay for . . . and I rather like my UAD 1176LN ;)
 
I don't see how the lookahead would cause many problems. A lot of plugins like compressors already use it.

And if it's just on the master buss I don't think latency would be an issue.

LAME is open source and is the most commonly used. Also, I think the DLL can be used to encode streaming data (which is really how WAV files are formatted anyway).

I could be wrong. Maybe it was a hopeless idea. :(
 
LogicDeLuxe said:
MP3 does no EQ'ing and no noise either. It works on omission of frequencies.
Click to expand...

To my understanding it's the upper and lower partials that are removed (I could be wrong on this). If so, would steep slope high and low pass filters get you in the ballpark?
 
Last I checked - which was maybe a year or two ago, it may have changed since then - the ISO codec (which is I believe what LAME is based upon) was public, however the Fraunhoffer codec was privately licensed only.

G.
 
You must log in or register to reply here.
Back
Top