16bit or wav have nothing to do with levels.
What he might be telling you is that the 16bit file he sent you was ment to be replicated on CD and you should have ask other version of your master for diffusion on other medias.
Online streaming, youtube, CD, tv and radio all call for different mastering approach. If he is a professional, that's what he ment.
ask him to clarify in words you can comprehend. I'm sure if you stay calm he will take the time.

Gazukmale, post: 449262, member: 50495 wrote: I'm really confused because i thought there was no difference in the 16 bit wav file whether it be for replication or to be converted/encoded by music stores to mp3?

I share the same thought.
16/44.1 is the standard for CD but as far as a 16bit file being too loud, I've never heard of that. Before you take my word on that, lets see what the other guys here say when they get done for the day.

pcrecord, post: 449263, member: 46460 wrote: 16bit or wav have nothing to do with levels.
What he might be telling you is that the 16bit file he sent you was ment to be replicated on CD and you should have ask other version of your master for diffusion on other medias.
Online streaming, youtube, CD, tv and radio all call for different mastering approach. If he is a professional, that's what he ment.
ask him to clarify in words you can comprehend. I'm sure if you stay calm he will take the time.

Thanks for reply
Well i talked to CD baby and they said they have never heard of what the mastering engineer was saying...
CD baby said that the 16 bit file shouldn't effect the levels weather it is for replication or not(but they said they could be wrong) .

Then I talked to the mastering engineer...
He said that the 16 bit wav file can only be used for CD replication as the levels will be too loud for digital use online.
Even if its converted to MP3 it will distort. I then asked if it was DDP . But i said that its not DDP file , its a wav. He then said its the same process as a DDP, but hes created a wav file.

As far as i was aware if something is a 16 bit wav file and is converted to MP3 or other formats, no distortion should occure only quality loss?
Now I'm totally lost? or am I missing something? If hes right , the lack of communication is very bad . And it seems strange as he was asking me how my release was going?

Gaz

audiokid, post: 449264, member: 1 wrote: I share the same thought.
16/44.1 is the standard for CD but as far as a 16bit file being too loud, I've never heard of that. Before you take my word on that, lets see what the other guys here say when they get done for the day.

Chris, aren't you getting that what the ME said wasn't hat 16bit was too loud but the specific file that he made was ? Of course I might be wrong and it's a fake ME ;)

Gazukmale, post: 449267, member: 50495 wrote: He said that the 16 bit wav file can only be used for CD replication as the levels will be too loud for digital use online.

Well that's none sens since he is the one deciding for the level he puts into the file !! LOL !!

pcrecord, post: 449268, member: 46460 wrote: Chris, aren't you getting that what the ME said wasn't hat 16bit was too loud but the specific file that he made was ? Of course I might be wrong and it's a fake ME ;)

It all sounds a bit weird.

pcrecord, post: 449269, member: 46460 wrote: Well that's none sens since he is the one deciding for the level he puts into the file !! LOL !!

Thats what I am thinking as well. Sonically a 24bit may sound a bit smoother but is shouldn't sound louder because its being used as a master to create a CD or MP3 for that matter..

pcrecord, post: 449269, member: 46460 wrote: Well that's none sens since he is the one deciding for the level he puts into the file !! LOL !!

Ok so now I asked the ME (to see if he was trying to rectify a mistake)...
"So the first file you sent me for replication is ok to for me to send to the CD replication company, as im having some CD's replicated next week.
And the next file I receive from yourself is to send out to CD baby again"
He answered Yes ...ummm ...
so maybe he is telling the truth. I guess I can always copy a CD on my computer to see?
I listened to my track (The 16 bit wav version apparently for CDreplication only) and there does seem to be some distortion in the sub bass half way through the mastered song that i cant hear on the mix. :-/ this isnt my day , especially when the song has gone out to most stores ..(b#llocks)

Gazukmale, post: 449277, member: 50495 wrote: Ok so now I asked the ME (to see if he was trying to rectify a mistake)...
"So the first file you sent me for replication is ok to for me to send to the CD replication company, as im having some CD's replicated next week.
And the next file I receive from yourself is to send out to CD baby again"
He answered Yes ...ummm ...
so maybe he is telling the truth. I guess I can always copy a CD on my computer to see?
I listened to my track (The 16 bit wav version apparently for CDreplication only) and there does seem to be some distortion in the sub bass half way through the mastered song that i cant hear on the mix. :-/ this isnt my day , especially when the song has gone out to most stores ..(b#llocks)

Let's assume he knows what he is doing(I'm trying to understand).
One tool most ME use is LUFS metering. This tool mesure and sum the quiet and louder parts of the song and give a better measurements. This means, if there is a long quiet part in the song and none in another song, if you put both at the same LUFS level, the one with quiet parts may get louder on its louder parts compared with another song with less dynamics.
A fact is, streaming sites have their level standards and when we upload to certain sites the level will be changed accordingly. Those sites mostly use LUFS as well.
So, If the ME is aware of all this, he might offer more than one master files to fit the level requirements of different diffusers / medias.

But !!! of all the digital formats, I never heard that one is louder than an other. OK maybe one can be pushed harder, is that it ??
In any case, if I export a song to different format, 24bit, 16bit, mp3 etc.. I'm sure they will be equally loud.

If you could invite him on the forum, I'm sure many of us members would like to know more about those things.. ;)

pcrecord, post: 449278, member: 46460 wrote: If you could invite him on the forum, I'm sure many of us members would like to know more about those things..

we'd likely need to edit this post and the title though. If that was a consideration, I would be happy to do the edits.

It's a real phenomenon - there have been a few threads and technical explanations over on sound on sound forums recently...I've not got time to search for them right now....I'll check later if my brain remembers.

Bottom line is that the MP3 conversion process can add intersample peaks that can result in audible distortion if the initial level of the source file is too high...so it's recommended that the peak level of any digital file that's going to be converted is left with 2 to 3 dB of headroom.

Keith Johnson, post: 449299, member: 49792 wrote: It's a real phenomenon - there have been a few threads and technical explanations over on sound on sound forums recently...I've not got time to search for them right now....I'll check later if my brain remembers.

Bottom line is that the MP3 conversion process can add intersample peaks that can result in audible distortion if the initial level of the source file is too high...so it's recommended that the peak level of any digital file that's going to be converted is left with 2 to 3 dB of headroom.

Thanks ..So that would mean that the file to be used for audio replication could peak when converted. That makes sense actually now.

The bit resolution itself has nothing to do with the amplitude level of the content. Now... The FORMAT can - as Keith (Keith Johnson ) mentioned - intersample peaking has been known to occur in the conversion from wave to MP3 ... but that's caused more by the codec "rounding" down or up; so Keith mentioning that a headroom of 2db -3db being allowed for that format is a good idea...

That being said...Probably what happened, is that the ME made one PCM 16 Bit (Redbook) master for CD replication, which - by and large - would have a LUFS level of around -12db, with peaks just below 0db.
Using this same LUFS level could prove to be problematic for other types of distribution, like internet streaming, or for MP3 download services like iTunes, where they have different LUFS level requirements. Some want a LUFS level of -16, while others want it as low as -23db (The EBU has an actual broadcast standard of this level, it's referred to as EBU128)
I don't think your M.E. was "conning" you... it just appears that he wasn't explaining himself very clearly to you as to why he did what he did.
It's not at all uncommon for there to be different masters for different uses, especially these days, when there are so many digital avenues in which to release and have your music heard and sold.
FWIW
-d

Keith Johnson, post: 449299, member: 49792 wrote: It's a real phenomenon - there have been a few threads and technical explanations over on sound on sound forums recently...I've not got time to search for them right now....I'll check later if my brain remembers.

Bottom line is that the MP3 conversion process can add intersample peaks that can result in audible distortion if the initial level of the source file is too high...so it's recommended that the peak level of any digital file that's going to be converted is left with 2 to 3 dB of headroom.

Yep, try it yourself. Lossy compression can shift peaks around, and some of those will be higher than the uncompressed file. Actually, you can do the same with eq. Normalize a mix file to -1dB, high pass it, then check the peak level. It will likely clip.

I'm not a professional ME and never want that job but have some pro gear and a fine monitor system to do mastering as a business. I only mention this so we get the drift that I'm not missing the point because of gear quality.

That being said, I've mastered hundreds of tracks for myself, friends, clients for the hell of it, for fun etc etc etc.. and have never had this problem once. I mean... I have never heard something magically appear online that I didn't recognize as audible ("ah, there it is :oops:") on my system. I have missed the cues many times as this is part of my learning and listening curve to improving and recognizing cause and effects. This is why Mastering Engineers need an excellent monitoring system and DAW that doesn't BS.

The only thing I can think of that creates this type of distortion in a file is when I have pushed level too hot that becomes more obvious after its been reduced to mp3. But its not like something blindly appears in the MP3 that you can't go back to your wave file and hear it there too.

I don't blame bad work on the MP3 or upload process. I know we blame online for many of the bad sonics but experience tells me that I missed it happening in my DAW. A good wave file never ends up distorting after its been uploaded if I did it right in the first place.

I think he missed it which happens to the best of us.

Something to look forward for in the next year or so, Anything that is being processed, bounced down, MP3 etc... an indicator (big red bell goes off!) when distortion is detected. These issues will all be gone one day. Kind of like autopilot. One day we won't have to even listen to the final mix lol. It will all be rendered and wrapped up all pretty with a bow around it. Artificial Intelligence.

I think he or someone missed it.

audiokid, post: 449308, member: 1 wrote: The only thing I can think of that creates this type of distortion in a file is when I have pushed level too hot that becomes more obvious after its been reduced to mp3. But its not like something blindly appears in the MP3 that you can't go back to your wave file and hear it there too.

That's the thing with the loudness war, some are still Under the impression louder is good. Pretty sad.
Thing is, this discussion made me Wonder if I won't change my pseudo mastering end levels.
I'm currently putting Pro-L with an output of -1db and pushing the level up to the LUFS reading of my choice (depending where the file will go).
I'm just thinking, should I put the max peaks to -3db ??
What do you think ?

Converting wav to mp3 changes peak values. Whether it's audible on a given system is a separate question.

bouldersound, post: 449311, member: 38959 wrote: Converting wav to mp3 changes peak values. Whether it's audible on a given system is a separate question.

So what you are saying is the peak values are added after the sum and ideally we should be listening and finalizing, mastering the MP3 (not mastering a 16bit wave) so we don't miss what is creating this peak one step back?

Realtime AAC/MP3 audition plug-in?

audiokid, post: 449312, member: 1 wrote: So what you are saying is the peak values are added after the sum and ideally we should be listening and finalizing, mastering the MP3 (not mastering a 16bit wave) so we don't miss what is creating this peak one step back?

Realtime AAC/MP3 audition plug-in?

How you handle it is up to you. As far as I know it's a matter of: render the wav file, convert to mp3, check for clipping, go back and lower the peaks in the wav file if needed until the mp3 doesn't clip. But there may be tools that simplify the process. A plugin that could tell in real time when something would clip after a given type of compression would be helpful, especially for situations where you're uploading a wav and letting the streaming site convert it. Otherwise it would be a time consuming process and metering the result would be a pain.

I'm curious. Would this happen in a mono file?

The last time I talked to an ME specifically about having separate MP3 and CD masters, he said he set the compression and limiting differently for the two formats. His argument was along the lines that, as CD players are increasingly less common in new cars, you need a version with less overall dynamic range for playing in vehicles. Given that he was dealing with my folk/jazz recordings and not rock or other styles that have inherently self-limited DR, I'm not sure I buy that argument.

bouldersound, post: 449311, member: 38959 wrote: Converting wav to mp3 changes peak values. Whether it's audible on a given system is a separate question.

I'm not trying to be an ass here but I have a desire to learn all I can by questions over and over until I get it right. I really want to get to the bottom of what you are saying here, that I simply do not believe (YET). :)

What I am hearing said over and over is no different to someone saying;
"The guitars were in tune or in phase, but now they are out of tune or out of phase because its an MP3 now."

Then my inner voice keeps saying this about lower bit rates, ...

Is this distortion actually revealing my lack of attention to detail? Is the MP3 code having trouble rendering the (L-C-R) adjustments and the result are the sound of a bit pile up of audio that ends up peaking into swirly distortion?

So far I am convinced this isn't the fault of the mp3. It is the indication of recording, mixing and mastering improvement. The MP3 is simply making it all more obvious.

Kind of like reducing soup until you taste all the salt in it.

Boswell, post: 449316, member: 29034 wrote: The last time I talked to an ME specifically about having separate MP3 and CD masters, he said he set the compression and limiting differently for the two formats. His argument was along the lines that, as CD players are increasingly less common in new cars, you need a version with less overall dynamic range for playing in vehicles. Given that he was dealing with my folk/jazz recordings and not rock or other styles that have inherently self-limited DR, I'm not sure I buy that argument.

I don't think the ME the OP was dealing with was talking about dynamics. He was saying that if the file is going to undergo psycho-acoustic (lossy) compression that the song's peak level would rise and that there had to be some accommodation for it. It seems to me it should just be built into the compression algorithm.

audiokid, post: 449317, member: 1 wrote: I'm not trying to be an ass here but I have a desire to learn all I can by questions over and over until I get it right. I really want to get to the bottom of what you are saying here, that I simply do not believe (YET). :)

I totally get that and appreciate your skepticism. Try it yourself. Render a wav file with 0dB peaks, compress it to mp3 and reimport it. I didn't believe it until I did the test myself.

bouldersound, post: 449319, member: 38959 wrote: I totally get that and appreciate your skepticism. Try it yourself. Render a wav file with 0dB peaks, compress it to mp3 and reimport it.

Does that happen for you with a mono file?

bouldersound, post: 449314, member: 38959 wrote: How you handle it is up to you. As far as I know it's a matter of: render the wav file, convert to mp3, check for clipping, go back and lower the peaks in the wav file if needed until the mp3 doesn't clip.

iZotope Ozone has a codec preview ment to help listening to what will happen to our files once converted..
I honestly didn't care about it until now.. but it's gonna be on my radar on my next projects..

https://www.izotope.com/en/community/blog/tips-tutorials/2015/11/using-codec-preview-in-ozone-7-advanced

audiokid, post: 449320, member: 1 wrote: Does that happen for you with a mono file?

Yes. I started with a stereo file with -0.3dBFS peaks, converted it to mono using only the left channel (no summing), confirmed it still had -0.3dBFS peaks, converted it to a mono mp3, reimported it. I can see multiple-sample sequences of 0dBFS.

In the image below the wav is on top, the mp3 is below. It's the same point in the song, though the time selection shows it slightly different due to mp3 compression adding a bit of delay at the start of the file. The wav is peaking at -0.3dBFS while the mp3 is peaking at 0dBFS. The wav is already somewhat flat-topped due to client demand for stupid loud, but I did leave that -0.3dBFS of headroom for the benefit of converters that lack the headroom to deal with ISP.

pcrecord, post: 449323, member: 46460 wrote: iZotope Ozone has a codec preview ment to help listening to what will happen to our files once converted..
I honestly didn't care about it until now.. but it's gonna be on my radar on my next projects..

Sequoia does this with flying colours. The next version will be improving upon it. iZotope and Sequoia are where I would be looking as well. And Fabfilter ProL is a staple for me. Use that and you never have a problem you can't hear.

Quite frankly, I never have this problem, never experience it. I simply do not relate to any of this but I understand it and believe its being noticed.

bouldersound, post: 449324, member: 38959 wrote: Yes. I started with a stereo file with -0.3dBFS peaks, converted it to mono using only the left channel (no summing), confirmed it still had -0.3dBFS peaks, converted it to a mono mp3, reimported it. I can see multiple-sample sequences of 0dBFS.

In the image below the wav is on top, the mp3 is below. It's the same point in the song, though the time selection shows it slightly different due to mp3 compression adding a bit of delay at the start of the file. The wav is peaking at -0.3dBFS while the mp3 is peaking at 0dBFS. The wav is already somewhat flat-topped due to client demand for stupid loud, but I did leave that -0.3dBFS of headroom for the benefit of converters that lack the headroom to deal with ISP.

exactly my point! But... I like where this is going now. I'm open to being schooled on this one.

Please try this with a mono wave file now. Not stereo. (mono to mono)

audiokid, post: 449327, member: 1 wrote: exactly my point! But... I like where this is going now. I'm open to being schooled on this one.

Please try this with a mono wave file now. Not stereo. (mono to mono)

I just did that. I discarded the right-channel information and converted the wav format to mono. It is in fact a mono wav file with -0.3dBFS peaks. You can see for yourself in the pic that there are two mono audio files, one wav and one mp3. That it came from a stereo file is irrelevant.

What you're suggesting is that it has something to do with altering of the stereo image by the compression algorithm. That is not the case here. I did not give the mp3 encoder any stereo image to mess with, just one channel of audio in wav format that it converted to one channel of audio in mp3 format.

bouldersound, post: 449329, member: 38959 wrote: I just did that. I discarded the right-channel information and converted the wav format to mono. It is in fact a mono wav file with -0.3dBFS peaks. You can see for yourself in the pic that there are two mono audio files, one wav and one mp3. That it came from a stereo file is irrelevant.

What you're suggesting is that it has something to do with altering of the stereo image by the compression algorithm. That is not the case here. I did not give the mp3 encoder any stereo image to mess with, just one channel of audio in wav format that it converted to one channel of audio in mp3 format.

Thanks you for doing this.
I am now wondering if ALL DAW's fall apart here.
Boulder, you are on Vegas, correct?

When I can actually get my new studio back, I am going to pursue this with Sequoia and will repost back with my results.

In the mean time, I think this deserves a bunch of DAW's being tested.

Gazukmale, post: 449277, member: 50495 wrote: I listened to my track (The 16 bit wav version apparently for CDreplication only) and there does seem to be some distortion in the sub bass half way through the mastered song that i cant hear on the mix.

Can you clear that sentence up?
Do you hear distortion or not, if so, what are you listening through for playback?
How reputable was the ME?

If there's distortion in the CD meant as the replication master that seems like a concern regardless of artifacts the mp3 conversion may have.

audiokid, post: 449330, member: 1 wrote: Thanks you for doing this.
I am now wondering if ALL DAW's fall apart here.
Boulder, you are on Vegas, correct?

When I can actually get my new studio back, I am going to pursue this with Sequoia and will repost back with my results.

In the mean time, I think this deserves a bunch of DAW's being tested.

I'm not using a DAW, I'm using Sound Forge 6 to take the right channel information out, change the wav format from stereo to mono and examine the files. It's ancient but that is irrelevant. It's not a matter of metering. I used the Statistics function to confirm the change in peak level from -0.3dBFS to 0dBFS that can be seen in the waveform display.

I just put the same two files into Audacity, and used Audacity to make a mono version of the original stereo wav (split track, delete right channel), and got the same results.

Are we defining and confirming this is what happens to all MP3 files no matter what DAW platform?
All DAW's use the same process to arrive at an MP3? I am aware their is a patent on MP3 but was not aware of how we arrive at it all the same way.

Summary: The least expensive MP3 process always does the math the same.

My next question:
Hypothetically: If we were to import a wave session done at example: Sterling Sound, only to use Sound Forge 6 to turn it into an MP3, it would end up with a change in peak level from -0.3dBFS to 0dBFS .. thus... same sounding MP3?

I didn't use Sound Forge to make the mp3. Generally speaking I don't make mp3 files myself, I upload wav files and let the site (e.g. to Soundcloud) make the conversion. So I had to search around to find something to convert the file. All I could find was Cool Edit that's been on this machine for at least a decade. Sound Forge is not a DAW, it's an editor, as is Audacity. Cool Edit is also an editor though it has some DAW-like capabilities.

It would be a good thing for others to repeat this with different editors and different mp3 encoders. I'm going to install LAME so Audacity can be used to convert wav to mp3 and try this again without using Sound Forge or Cool Edit at all. My money says I get the same result regardless of the editor or encoder.

audiokid, post: 449335, member: 1 wrote: Are we defining and confirming this is what happens to all MP3 files no matter what DAW platform?
All DAW's use the same process to arrive at an MP3? I am aware their is a patent on MP3 but was not aware of how we arrive at it all the same way.

Summary: The least expensive MP3 process always does the math the same.

My next question:
Hypothetically: If we were to import a wave session done at example: Sterling Sound, only to use Sound Forge 6 to turn it into an MP3, it would end up with a change in peak level from -0.3dBFS to 0dBFS .. thus... same sounding MP3?

From what I understand MP3 conversion in general, reguardless of brand, can cause/reveal inter sample peaks/distortions, and does come out slightly hotter than the .wav. I recently read an article somewhere, or maybe on the tube, where the engineer was saying they leave more headroom on a separate mix destined for MP3 conversion. Like -.5 or -1dbfs or something.

Whether or not all conversion algorithms are created equally or not I'm skeptical, but from what i understand they all do those things to some extent.

I'm excited to compare Samplitude, sequoia, izotope, and sonnox codecs to hear it there's any audible differences or not. Codec shootout anyone?

Whats interesting, when I am exporting a mix to MP3, or online for that matter> letting the online software do their compression, I never hear my online version distort. In Sequoia its so simple. I set my limiter to -01 (a bit below 0) and it sounds like it should online.

So because I don't hear my mix distort, I have never looked to see this happens. I'm surprised actually. Ill do some shootouts once I get new converters.

I also wonder if the source sample rate or bit depth has anything to do with it.? Like if you convert to MP3 from say 24/96 or from 16/44.1. I wonder if there's any general rule of thumb on that?

Thats interesting. I mixdown/ finalize to a second DAW as you know, which is always 44.1/24 . Boulder, are you doing these comparisons at 44.1 ""?
I think he said he was, Kyle.

All done at 44.1kHz/16 bit.

Many modern converters have headroom in the analog parts of the circuit to accommodate intersample peaks, though I think that's technically a violation of the Redbook standard.