Bringing a master back in to Sonar can mess people up, playing the mix through the same FX by accident.

 
Not a problem, so long as you set the output of the track directly to Main Outs. I usually bounce the Master bus to a "Master" track in the project pointed to Main Outs, and use that for subsequent exports to MP3, 44.1 x 16 WAVs, etc.
 
 
 
 

The problem may be that you're setting the limiter to 0db or very close to it. When encoding to MP3 you need to leave a little more headroom than when exporting to PCM, to prevent intersample clipping caused by ringing in the steep filters used in the encoding process. Try setting your brickwall limiter to -2db or even -3db. You won't notice any difference when the file is broadcast, but that should avoid the distortion.

exporting earlier with the "entire mix" button selected and indeed there was a 6db amplification going on causing the majority of my volume problems... but what wonders me even more, why didn't these clippings and distortions happen in the past ?

 
You must now have two buses wih the same mix running in parallel to different outputs (e.g. a headphone mix as Jonesy mentioned) that you never had in the past...? Maybe a send on the Master to a headphone bus/out? When you play back, you only hear the Master bus, but when you export Entire Mix, you get both outputs.

Many are referring to the OP. We are actually solving the problem of a hijacker. The OP "cpybee"
is long gone.

What we need a clear answer to now is, as others said:

-Does the problem occur in a similar way when the export to wav is committed before converting to MP3
- does a little more headroom affect the results
- So many doubts are pointed to exporting "entire mix" that I'd recommend testing with other options, like "master".

Many are referring to the OP. We are actually solving the problem of a hijacker. The OP "cpybee"
is long gone.

There are open questions

-Does the problem occur in a similar way when the export to wav is committed before converting to MP3
- does a little more headroom affect the results, as Bitflipper suggested
- If you've played with the Lame settings, is everything OK there

So many doubts are pointed to exporting "entire mix" that I'd recommend testing with other options, like "master".

"the levels that i see at the mixdown now, are still not the same in the exported file.."

If I have understood things right you can never have the same levels in the mixdown as in the direct MP3 export, because MP3 conversion tampers with the levels, it raises them. It's been mentioned half a dozen times here already.

it even distorts without using any plugins or limiting with a moderate soundlevel that peaks at max -3db and on the wav file... so it's not an mp3 issue

So you get the same distortion when you export it as a wave?

I'm in agreement with Randy Bowser, I would never export direct from Sonar to MP3, I always export to WAV file at 44/16, then fire up GoldWave so I can increase the volume level of the WAV file if necessary, then export to MP3.

the wavein the screen shot looks like it was normalized...

You can import the MP3 back into the project and A/B it with the original, but it's hard to objectively compare an MP3 to the original track because it's no longer the same file. It's been pretty badly mangled, from a data point of view.

No, we can't attach files on this forum, but you can upload a sample to some online file depository and link it.

I'm trying to imagine what "underwater" sounds like. Others have actually used the same term in the past, but I still don't know what it means. Aliasing, maybe? See if you can find a way to post a sample. Better yet, post both the original wave and the MP3.

PLEASE do yourself a favor and switch to recording in 24-bit. The fact that most DAW programs STILL default to 16-bit recording in an age of Terabyte hard drives is utterly ridiculous. By recording at 16-bit you are getting less than full CD resolution the moment you are quieter than just below clipping. If you record at 24-bit you can record with peaks at -12 dB and still have plenty of resolution to get quality recordings even at very low levels, resulting in better quality when processing and truncating to 16-bit with proper dither for CD masters.

In my personal and not so humble opinion, recording tracks at 16-bit is audicide.

To quote Wikipedia on the matter-

"The resolution of a 16-bit system can be calculated by using 2¹⁶ which gives a value of 65,536. A 24 bit system (2²⁴) has a resolution of 16,777,216."

AND

"24-bit digital audio has a theoretical maximum dynamic range of 144 dB, compared to 96 dB for 16-bit; however, current digital audio converter technology is limited to dynamic ranges of about 120 dB (20-bit) because of 'real world' limitations in integrated circuit design."

Peace,

Tunes

BluesMeister I think you're largely correct here, especially in practice. Most people don't record at a bit rate higher than 24-bit (unless we're talking a soft synth or internal processing there is no reason to do so, as most converters don't really capture useful information more than about 20 bits due to limitations of design and basic physics.) So recording at a higher bit depth than 16 usually means 24, which is takes only 1.5x the space.

Most folks that tend to up their sample rate tend to go for 96 kHz or above, which would take more than 2x the space for CD quality (44.1kHz.) There are a few folks that prefer 88.2kHz to 96kHz as the math to CD quality is a bit more linear, but at the end of the day it all depends on the quality of your converters/conversion software and the filters they use to cut out frequencies that are higher than they can capture. For what it's worth in that regard I've found Voxengo's R8brain to be a great sounding converter.

Lastly, I'm sure you'd be fascinated to check out 1-bit digital recording. In a lot of ways it is simpler than the way we're used to thinking. It's not broadly used yet, but is as close to analog as digital has ever come.

http://www.korg.com/services/products/mr/Future_Proof_Recording_Explained.pdf

Most of our modern converters already use a clock that runs at 2.8 or 5.6 MHz and use circuitry to subdivide this signal according to the sample rate we specify. With 1-bit recording that circuitry is removed and recording is done at the original clock speed, making the effective sample rate approx. 2.8 million or 5.6 million Hz!!! Instead of using 16-bits or 24-bits to record the values, like an x/y graph, in 1-bit you basically record the answer to a simple question 2.8 or 5.6 million times a second: is the sound getting louder (storing a 1) or softer (storing a 0.) This is the same DSD technology used in SACD recording.

http://en.wikipedia.org/wiki/Direct_Stream_Digital

Peace,

Tunes

Bit, is it a linear increase in file size? I used the term exponential in a perjorative sense. I've only ever tried recording at 96kHz once to see what would happen and the file size was huge compared with a similar recording at 44.1kHz so I never bothered with 96 again.

This is a most interesting thread, Op asked about distorted export, thinks he solved it by turning off dithering???
Then someone is selling clothing??
Then thread is high jacked by a guy who people try to help and he disappears with out a word of resolve??? ( Rude?)
Then a 3rd person high jacks the thread asks a question and disappears??
And now we are in an interesting discussion regarding file size  that started because post #49 is talking to god only knows who about 16 bit recording??  Is he talking to the OP who left the form back in November? Certainly not the rude radio guy? who has also long gone?
Anyhow please let this one fade away :)

BluesMeister


Linear fade, slow fade or fast fade?

--or my favorite - a nicely sculpted exponential fade with a few trick momentary fade ups and back down again before totally going sloooooowly out.

RB