Exporting music question

[echoshock]

So a back to basics question. Its frustrating when I finish mixing a song, export it to an mp3 file and realize its basically half the volume of any other commercial mp3 file I have.

I'm working in Sonar at 24 bit 88khz. I'm then exporting to an mp3 file at 16 bit and 44.1. I guess the conversion from 24 to 16 is most of the cause of the fidelity and volume loss. Should I be exporting to mp3 in some other way or bit depths?

Thanks

[UstadKhanAli]

Concerning loudness, I believe compression and gain optimization is your answer. That or simply turning your volume knob a bit farther to the right.

 

As for the fidelity issue, a higher-resolution MP3 if you must export to MP3 might create a little more happiness for you.

[echoshock]

I guess that is part of the confusion. My mix is about as close to clipping as it can be through limiting and compression etc, but yet the mp3 is still half the volume of anything else I hear. I physically can't make the mix any louder. Is 16 bit 44.1 the standard format for mp3s?

[Phait]

Have you tried converting the MP3 with other software instead of exporting? Are you using Constant Bit Rate instead of Variable Bit Rate?

 

"available sampling frequencies are 32, 44.1 and 48 kHz. A sample rate of 44.1 kHz is almost always used, because this is also used for CD audio" http://en.wikipedia.org/wiki/MP3#Audio_quality

[echoshock]

Thats a good question. If I have the mixed stereo file in Sonar, what is the best way to convert to mp3? I had thought the only way was to export from sonar.

I don't see an option to choose constant or variable bit rate. Which is preferred?

Thanks

 

Have you tried converting the MP3 with other software instead of exporting? Are you using Constant Bit Rate instead of Variable Bit Rate? "available sampling frequencies are 32, 44.1 and 48 kHz. A sample rate of 44.1 kHz is almost always used, because this is also used for CD audio" http://en.wikipedia.org/wiki/MP3#Audio_quality

 

[blue2blue]

echo

 

The loudness problem you're facing is not one of maximum volume but overall average volume.

 

If you google "loudness wars" or you'll eventually find more than you ever wanted to know about the issue. It's become an issue that's spilled out into the non-musician/non-engineer general public as the average level of commercial releases has been increasingly jacked up by using program compression/limiting. The process of increasing the average volume until there is very little dynamic range [the ratio between the loudest and quietest bits is reduced until almost the entire tune, what was once quiet as well as loud, is all made loud]. This process has become known as "squashing."

 

 

An average volume (sometimes called RMS average, for root mean square, a system of averaging over the duration of a given section) of -16 to -14 dB used to be a 'normal' RMS average for pop music.

 

As long as people listened to albums, they usually didn't notice the jump in average level from one record to another. But as CD changers came into use -- and particularly as Mp3 players came into use and people began mixing tracks up from a wide variety of albums -- the jumps in level became increasingly apparent.

 

The ear, as you may know is a "cheap date." Throw a loud sound at it and a quiet sound -- and the ear almost always thinks the loud sound sounds 'better' (up to just under the point of discomfort, anyhow).

 

This has been well understood for a long time but it wasn't much of an issue until people started sequeing between tracks form different albums/sources.

 

(Back in the era of 45 rpm singles, there was a very similar 'loudness war' that got so out of hand that by the end of the 60s -- and the decline of the importance of singles -- most singles were so squashed that they sounded like utter crap. When I was a teenager getting into audio, I was, for a short time confused when I realized 45 rpm records should be capable of higher fi -- yet often sounded awful compared to album versions of the same track [at least as long as you turned up the volume on the LP version].)

 

 

Anyhow, if you're comparing your track to today's mainstream rock records, you may actually be trying to 'compete' with music with an RMS average as high as -8 dB. There have been reports of even 'louder' tracks.

 

At these compression levels, high frequency detail is typically lost, 'forcing' mastering engineers to crank the treble even as they squeeze the treble out with compression -- making the tracks sound increasingly harsh. (Google the title of the latest Metallica record and the word "squashed" and look at the complainst you'll find at BBs like this and GearSlutz.)

 

 

Anyhow, to the extent that you want to compete, you'll need to squeeze your tracks with compression/limiting, reducing the dynamic range from where it is (likely in the -16 to -14 dB RMS range) to something more 'competitive.'

 

Even so, this can be difficult, since bass energy can take up so much of that dynamic range.

 

So MEs trying to compete often roll the sub bass out and/or use EQ-band dependent compression in order to 'make room' for more of everything else -- increasing apparent volume even as it reduces/changes the quality of the basic mix.

 

_______________

 

With regard to apparent fidelity... there are a couple of issues that may be at work (independent of volume issues). The most likely, as others note above may be less than optimal settings for your mp3 encoder -- or an inferior codec (most of the worst codecs are no longer out there). Many folks like the LAME codec best -- but it, as well as the more common Fraunhofer codec -- have a wide range of settings (not all codec wrappers allow access to all the settings which is why some folks prefer the command line versions; recent versions of Sonar allow you to create your own 'wrapper' with different command line defaults in their codec manager). Either of these codecs can do a good job -- but only at slower settings with the right parameter settings.

 

Another issue that may be at work are potentially significant degradations you move from 88.2/24 bit to 44.1/16. While a well optimized 44/16 file should be capable of sounding almost indistinguishable from the higher resolution, stuff happens. Tools are imperfect -- and mixes and optimizations can be, as well; 44.1/16 at its very best allows high quality -- but it's easy to come up with less than the best.

 

Now, the latest versions of Sonar have pretty good sample rate conversion algorithms -- but earlier versions of Sonar included lower quality SRC software -- and that potentially may be combining with other factors to create more noticeable degradation.

 

Here's a website that offers test results for a number of sample rate converters: http://src.infinitewave.ca/

[UstadKhanAli]

The maximum vs. overall average volume is precisely why I mentioned using a combination of compression and gain optimization.

 

Now, I don't like to get too hung up on this stuff because I personally don't care whether my mixes are loud or not unless the client really wants it so. But if you must, run it through a couple of compressors and then a gain optimization plug-in or two and that should do it.

 

Also, for clarification, are you saying that the MP3 mix that you are exporting is half the volume of a WAV or AIFF file of the same mix that you are exporting?

 

BTW, I use iTunes to convert my AIFF files to MP3, not my DAW.

[echoshock]

 

Also, for clarification, are you saying that the MP3 mix that you are exporting is half the volume of a WAV or AIFF file of the same mix that you are exporting? BTW, I use iTunes to convert my AIFF files to MP3, not my DAW.

 

hmm I had thought this was the issue, but I tested again and the wav and mp3 appear to be the same volume when I import them back into sonar they are the same.

hmm now I'm confused lol.

Could someone maybe just describe their process from stereo 24 bit file in your daw to mp3? What process do you go through and what settings do you use to convert to mp3.

Yes, I'm aware of the loudness wars. But my mp3s are soo freakin quiet they sound like nothing. Like I mentioned, just to see how loud I can get them, I've used limiting to squash them as loud as possible but its the same thing. In other words, in my daw it seems they cannot be any louder without clipping.

[blue2blue]

I know it's long but did you read my post, which addresses loudness as well as SRC issues?

[echoshock]

 

I know it's long but did you read my post, which addresses loudness as well as SRC issues?

 

wow blue sorry. I actually read your entire post up to where you inserted the line break. I literally never even saw the next section you wrote.

Thanks for your info. Let me go back and read it again.

BTW: I'm using Sonar 7

[blue2blue]

Hey, no prob... I tend to go on...

 

Sadly... even with my longwindedness, it's probably not possible to address all the issues that may be contributing.

 

It's likely there are several contributing factors or overlapping issues.

 

 

What kind of settings are you using when you export from Sonar? (Which I also use, although I'm back on Sonar Pro 6.)

 

For reference, here's the export dialog page from Sonar 6:

 

Your dialog box may be a little different on Sonar 7, of course.

 

One thing to watch for are using "Fast Bounce" for finished mixes. Fast bounce can have a less than ideal-to-really bad effect on some soft synths. It messes with BFD on my system, as well as the Edirol GM synth that came with some version of Sonar or other that I have, it messes with the Perfect Space convolution reverb that comes with Producer, etc, but generally has no apparent effect on straight wave files in my experience. But I avoid it for any finished work -- and since it is a parameter switch in track and synth freezes, I tend to avoid it there except for quick checks and such.

 

Another issue is the type of dither (extremely low level shaped noise added to introduce randomness at the least significant bit to eliminate so-called truncation error) -- but this isn't likely to make any sort of huge difference. As a rule, Sonar will add dither when it's appropriate in its internal operations, but you do have a choice when it comes to exporting. But the dither offered in Sonar is well respected and the differences are subtle. You can read about the differences between the differnt types in the Help files.

 

Some folks will try to add dither using plug ins where it is not needed and that can add noise if done inappropriately; but, again, that doesn't sound like the problem here.

[echoshock]

I believe I was using fast bounce. Actually all of those boxes were checked off.

Yea, I understand there are probably a lot more items responsible also including my mix.

 

I just tried using exporting with Fast Bounce turned off and Sonar keeps stopping as soon as the audio plays, giving me the 'audio stopped' message.

[blue2blue]

I wish it was easier to zero in on the issue.

 

I don't usually say this but -- how about posting a link to a clip? (One of those compromised mp3s would be fine.)

 

Of course, posting a clip of the finished product doesn't let us compare it to the original project in the DAW, but it might tell us something. (If you're shy you could email it to me, if you like. PM me and I'll send you my email address, if you want.)

[Jon Gnash]

 

So a back to basics question. Its frustrating when I finish mixing a song, export it to an mp3 file and realize its basically half the volume of any other commercial mp3 file I have. I'm working in Sonar at 24 bit 88khz. I'm then exporting to an mp3 file at 16 bit and 44.1. I guess the conversion from 24 to 16 is most of the cause of the fidelity and volume loss. Should I be exporting to mp3 in some other way or bit depths? Thanks

Restore the commercial mp3 to PCM, and restore your mp3 to PCM and compare them in a waveform editor. I bet your file will have many more transient peaks.

 

Right?

 

Those transients are what

[Jon Gnash]

Change your MP3 encoder. You might get lucky.

 

Believe it or not it

[Anderton]

Are you sure your master level isn't down when you export?

[blue2blue]

 

Restore the commercial mp3 to PCM, and restore your mp3 to PCM and compare them in a waveform editor. I bet your file will have many more transient peaks.   Right?   Those transients are what

[the stranger]

1. Set master fader at 0.

 

2. Set all tracks so the master is peaking at -6dB. (If you have a good balanced mix, select all tracks and adjust their volume together. One thing to remember is when the tracks hit the mix bus, if you have applied a compressor/limiter to the mix bus, this is going to probably dictate a need to go back and adjust the tracks levels. Compression will affect the balance of your mix. Some frequencies may be highlighted and need to be taken down in level slightly.)

 

3. Apply a compressor/limiter to the mix bus. Adjust settings until you bring up the perceived volume to where you want it. (Using an mp3 you are referring to will help you achieve the volume you want.)

 

An mp3 encoder should not be causing a drop in volume. It takes a lot of practice and using compression/limiting to make it work for you in a transparent manner, but it's worth the time and effort. Study how to properly use a combination of proper gain staging and compression/gain reduction to achieve the volume you are looking for.

[the stranger]

Also, an exercise that can really help understanding compression is to open up various wave files in your audio editor and try applying various levels of compression, studying the results. Yes, your ears are the best judge, but you can learn to understand compression by studing the visual effects on a waveform.

 

I had read a million articles detailing this subject, but it took the above type of experimentation to finally understand it. The visual aspects can help you understand how this process works, just don't trust your eyes to use it. Trust your ears.

[blue2blue]

Experimentation and practice is a great idea -- it's pretty much how I really learned everything (though I went through a formal recording program at a community college and took the core classes at another -- in those days I'd do anything for free studio time. ) You read or hear something and then you try it out in various ways, poking and prodding and keeping track of the results... sort of an informal version of the scientific method...

[the stranger]

Hope he got all this.