MP3s To Audio CDs, opinion on getting best audible result Options

[d2e]

(Please excuse any ignorance as I'm not well versed with this and learning.)

I'd like to get started with converting my MP3s to audio CDs. The CDs I want to create will be used for CD-style DJing. I was given some advice on this conversion process to have my end result CDs sound really good.

The advice is below. If anyone wants to add any additional advice or feedback on this process, I would appreciate it! I have to know that there's no one way to doing this.

I'll also mention that some of my MP3 are albums, many are not and even some of my MP3s that are a part of an album have different bit rates. So, I want to take this all into consideration in doing this conversion.

• Use MP3Gain on my mp3s if clipping is detected.
  Specific settings weren't discussed but I have started testing the options in MP3Gain tonight to get my feet wet.
  
  
• Consider changing all mp3s to the same audible volume, which is a better option than normalizing.
  Specific settings weren't discussed on doing this.
  
  
• Decode using Foobar2000's diskwriter with 16-bit Dithered Output to .wav.
  This I'm told is the best decoding quality I can get.
  
  
• Burn my .wavs to good quality audio CD-Rs.
  My preference is to use Verbatim Vinyl CD-Rs. This would be a first for me using them and if they don't work well or are not suggested, I'd like to us Memorex CD-Rs as they HAVE worked very well for me. Preferred burning software is Nero 5.5.9.17.

Thanks...

d2e

[atici]

You will probably want the cue sheet if you have the albums. So that the position of the wave files on the cd will be exactly the same as the original. I guess even if you use a lossy encoder CDDB/freedb should recognize your CDs then.

[kotrtim]

Just use mediajukebox
apply Radio Gain on all your mp3s with MJ and burn them directly to CDs.
Isn't that faster & easier?

[kotrtim]

I guess even if you use a lossy encoder CDDB/freedb should recognize your CDs then.

I don't think so... only Windows Media db could do that.....

[d2e]

Atici, you're right about the cue sheets and for most of my albums thus far, I have this info.

Kotrtim, hmmmm... I've never used mediajukebox but I'm game for checking it out if it will give me good audio CD quality. But, I thought that first decoding .mp3 to .wav would increase my MP3 song quality prior to burning it to CD-R. No?

d2e

[atici]

The CDs I want to create will be used for CD-style DJing

Hmm may I ask why exactly you need MP3->WAV cds? Isn't it a better idea to keep MP3s as MP3s on CDs? I never use audio cds anymore so I have trouble understanding why Archive it once (even in lossless) and keep them so.

Edit: I have another idea. If you have 700MB discs maybe you can have multisession CDs, in which you can keep the audio CD in the first session and then the mp3 files as a CD-ROM in the second session. Considering most CDs don't have 600MB of information I guess that is a better solution because anyone that needs the mp3 files could have it without a transcode. I don't know how it would effect the compatibility though.

This post has been edited by atici: May 11 2003, 04:50

[kotrtim]

But, I thought that first decoding .mp3 to .wav would increase my MP3 song quality prior to burning it to CD-R.  No?

MediaJukebox's CD Burner on-the-fly is different from Nero's
it will create a folder ,decode .mp3 to.wav then burn it to CDs then delete the .wav file after burn completes

http://mp3decoders.mp3-tech.org/conclusion.html

This post has been edited by kotrtim: May 11 2003, 04:50

[Andavari]

But, I thought that first decoding .mp3 to .wav would increase my MP3 song quality prior to burning it to CD-R.

Decoding an mp3 to wav doesn't increase quality.

[d2e]

Atici - Well, going on the initial advice, I was under the assumption that the reason for going from MP3-WAV-audio CD was to give me a better sound quality than just using the MP3s as they are. But, if that's not so, then I won't consider that process.

The other option you mentioned seems like having the best of both worlds...I'll need to find out if the CD players I'm planning to use would support such a multisession CD. In thinking about this option, would I be able to fit an "average" CD of songs (let's say, 12 songs, 5:00 min each) in one session and its corresponding MP3s in the other?

===========

Kotrtim - I'll download it and check it out, thanks for the link. I'm curious though - in what way is MJB on-the-fly different from Nero's?

===========

Andavari - I'm glad you pointed this out. Let's say I go thru with this process anyway to create audio CDs (as a preference vs. just using the MP3s as-is on CDs).

Would I lose MP3 quality doing this? If I do, then, what process would be suggested if I don't want to use MP3s with the CD players and prefer to use audio CDs? Still keep in mind that I want good audio quality.

thanks...

d2e

[tigre]

...
I'd like to get started with converting my MP3s to audio CDs.  The CDs I want to create will be used for CD-style DJing.  I was given some advice on this conversion process to have my end result CDs sound really good.

The advice is below. 
...

FYI: The 'advice' d2e refers to was given here: http://forums.afterdawn.com/thread_view.cfm/33533

@ atici & Andavari: Ideally mp3 and audio CD-R made from mp3 sound identical, but there are some possible problems I mentioned in the other thread:

Theoretically mp3s and CDs made from mp3s sound identical. But there are a few issues ...

- On music recorded on high level lossy encoding (e.g. mp3) can cause clipping that occurs in decoding step. If there is clipping in an mp3 file you can check using aforementioned mp3gain. Mp3gain can change the volume of mp3 files without any loss to avoid clipping. You might want to use this for all your mp3 files, no matter if you plan to play them back directly or burn them to audio CD-R.

- You never know how good the decoder of a hardwareplayer is. Possible problems I know of:
-- incompatibility with VBR, certain bitrates (e.g. 320kbps, <128kbps)
-- no decoding of high frequencies (>16kHz) or distortion/aliassing at those frequencies)
-- Truncation instead of dither used
Of course you can test the player you want to use to find out if it plays everything fine, but with converting mp3 to audio CD-R you have controll over these issues by using decent software for decoding.

- In theory mp3 data CDs have better error correction than audio CDs as data CDs contain an additional layer of error correcton information. Decent audio-CD players can correct/interpolate errors (e.g. caused by scratches) and make them inaudible while uncorrectable errors in mp3 data CDs will lead to annoying pops/glitches/ skips.

@ d2e: Excuse me, if what I wrote has made you think:

But, I thought that first decoding .mp3 to .wav would increase my MP3 song quality prior to burning it to CD-R.

Let me try to clarify: If done properly, decoding to audio CD-R preserves 100% of the mp3 quality. If decent equipment used (and the mp3s are not clipping), playing back the mp3s directly from data CD-R will give you 100% quality too. IMO you'll just have a little bit more security and controll/choice what's happening if you use audio CD-R.

This post has been edited by tigre: May 11 2003, 15:49

[sxz]

@d2e

About audio quality from mp3


I tested Winamp2 & 3, MAD plug-in,and a lot of programs before Foobar2000.

After many tests I think that the best way to get a good WAV from MP3 is decode it with 32bit floating-point precision as Foobar2000 does. Most of soundcards have 16 o 24 bits precision, so the resolution must be changed obyously.
I have tested several dithers with diferent noise shaping (Cool Edit Pro 2.1, Wavelab dither, Foobar2000 with strong and soft noise shapening, etc) and the best by far is Waves L2-Ultramaximizer directX plug-in (I use this with WaveLab 4.0), and then I burn it with Nero.
The sound is clean and professional.

[d2e]

@ d2e: Excuse me, if what I wrote has made you think:

But, I thought that first decoding .mp3 to .wav would increase my MP3 song quality prior to burning it to CD-R.

Let me try to clarify: If done properly, decoding to audio CD-R preserves 100% of the mp3 quality. If decent equipment used (and the mp3s are not clipping), playing back the mp3s directly from data CD-R will give you 100% quality too. IMO you'll just have a little bit more security and controll/choice what's happening if you use audio CD-R.

hey tigre!

I made my way over here as you can see (and as suggested) and appreciate your carry-over help. Yes, from what you initially advised, I did think that the conversion of MP3->WAV->CD-R increased my MP3 sound quality. But, now I know it "preserves" it.

sxz - Thanks for chiming in also. I was able to follow you up until you said:

I have tested several dithers with diferent noise shaping (Cool Edit Pro 2.1, Wavelab dither, Foobar2000 with strong and soft noise shapening, etc) and the best by far is Waves L2-Ultramaximizer directX plug-in (I use this with WaveLab 4.0)

I'll need to read about dithering and what it does.

d2e

[atici]

sxz: Well if I had $1000 to spend on "Waves L2-Ultramaximizer directX plug-in" and "WaveLab 4.0" I guess I would not be using CDs anymore I'd look for 96kHz 24bit original records of my music

d2e: Yes you'll have enough space if you have both audio and data on the same CD. The 70 min audio cd as we know is 650MB. Most of the CD-R/CD-RW you can buy has 700MB space and considering most of your albums is shorter than 70 min, you'll have around 100MB free which is plenty for the MP3 files.

r2d2: What the hell are you doin' here buddy?

This post has been edited by atici: May 11 2003, 17:00

[KikeG]

To use dither in this context (decoding) is good, because it fixes requantization problems that happen when you decode lossy formats to 16 bits.

However, in practice, dither adds little to music perception, because it just fixes things at the 16th bit level, that is, things nearly inaudible or just inaudible at usual listening levels.

As to types of dither, in my opinion FB2k strong ATH dither is among the best availables when it comes just to playing music. Still, even if you use a more simple type of dither, such as the flat or triangular types of FB2K, you won't notice a difference in practice, because the audible differences here are even of less importance than the ones I talked before.

This post has been edited by KikeG: May 11 2003, 17:07

[sxz]

@d2e
Sincerely, I don´t know the algorithm of diferents dithers,
(here you can see the "static" noise type:

http://careta.4t.com

but when I play the CD through digital out conector of DVD to my Onkyo receiever it sounds as professional as CD.

If you want to probe: run WaveLab "Batch procesor" and add "Waves L2" with preset "Hi Res CD Master", add 32bit floating-point foobar2000 decoded file, and as output select WAV , 16 bit(for future add it to a preset).

@atici
I don´t have 1000$ neither, but according to Foobar200 Forum rule #9 we can´t talk about this.

This post has been edited by sxz: May 16 2003, 08:00

[d2e]

Atici: Thanks for the clarification of the multisession CD. Again, this sounds like a really good option that I very well may go forth with.

Kike: That breakdown helped me understand a bit better, I appreciate it!

Sxz: I'll take a look at that link, thanks.

In going thru with the initial suggestion, do I need to change any other settings in FB2k besides PCM 16-bit dithered in the Diskwriter section? I also noticed "ReplayGain mode"...do I want to use that here or just when using MP3Gain when I first fix any clipping in my MP3s?

Kike, that setting you mentioned appears to be set by default in FB, so I'll keep it as is.

d2e

[DickD]

This is a long post, but it's hopefully comprehensive and correct.

There are two kinds of ReplayGain:

1) Radio Gain (also called Track Gain)
This makes every track sound as loud as the next track, much the way things usually sound on the radio.

2) Album Gain
This makes every album sound as loud as every other album, but each track retains its relative loudness compared to the other tracks on the same album (so unlike Radio Gain, a quiet solo flute track does not get made as loud as a full-orchestra track on the same album)

Both methods work well for shuffle-playing and DJing. I prefer to use Album Gain (unless I'm given a compilation by a friend that wasn't replaygained, in which case I'll fix it using Track Gain, and might even copy the track gain values to the album gain field in FB2K). For professionally made compilation albums, however, I trust Album Gain.

In mp3gain, the MP3 itself is modified (each frame has a 'global gain' volume setting), so you can only choose one type of ReplayGain to apply to your files. This is often worth doing (e.g. Album Gain) as it will be volume-adjusted in every MP3 player. (MP3's global gain varies in 1.5 dB steps, which is accurate enough for practical purposes) No audio information is lost in the process, so it's completely reversible (if you enable the mp3gain logs and are prepared to reverse it by manually)

Foobar2000 can scan for Replaygain (right click menu) including scanning a set of files as an album. The Track Gain, Album Gain and peak-levels for track and album (to prevent clipping) are stored as values in the FB2K database (if enabled) and in tags on the files. This allows the FB2K user to choose Radio or Album gain (or no change) at time of playback (or time of diskwriting), and to choose whether to use clipping prevention, and even to play back at a different target loudness (other than 89.0 dB or whatever was chosen in mp3gain).

Note that if you apply mp3gain changes after FB2K, you need to re-scan the files, because the latest version of mp3gain doesn't know that FB2K has already applied tags to them (future versions are going to, I hear).

If you've already used mp3gain on, say the Album Gain setting, you'll notice that Foobar2000 will usually show Album Gain values within +/-1.5 dB (the step size used by mp3gain), but many of the Track Gain values could be outside this range.

Because most albums are mastered louder than necessary, Replaygain usually results in slightly reduced volume for modern albums and little change for classical and older albums.

The strong noise shaped dither (recommended) in the latest FB2K is similar to what professional CD mastering uses when converting from 20/24-bit to 16-bit for CD and makes the dither hiss (already very quiet) even quieter to the human ear, and reputable mastering engineers such as Bob Katz reckon this extends the effective dynamic range of CD to around 115 dB (similar to 20-bit with flat dither). This dither helps preserve the incredibly quiet parts that Replaygain will have made even quieter, that will usually only be audible in the fade-out when the volume is turned up really loud. It also prevents "truncation distortion" caused when rounding errors accumulate to cause tones related to the music (but not nice-sounding). This is really difficult to hear on 16-bit audio at normal volume, but it's better to be safe than sorry (especially if playing a fade-out in a loud club at the end of the night).

It's the dither and replaygain support in the same package that would make me use Foobar2000 in preference to other packages that can decode the MP3s and burn the CD in one.

If I were preparing a compilation CD using FB2K's Convert (a.k.a. Diskwriter) on the right-click menu, I'd use:

WAV (PCM 16bit dithered)
Use ReplayGain
Album Gain (with clipping prevention, and pre-amp set to 89.0 dB)
Dither (strong noise shaped)

Normally, I'd disable DSPs to avoid any accidental effects.

However, if I had any non-44.1 kHz files (such as 32 kHz or 48 kHz sample rates) I'd need to convert to 44.1 kHz so I'd also turn on Enable DSP in the diskwriter settings and enable Resampler (SSRC), setting it to 44100 S/s, Slow mode, 64bit for best quality. This will do nothing to files that are already 44.1 kHz.

Also, if I had any mono files, CD needs stereo, so I'd use the DSP called 'Convert mono to stereo' (i.e. same on each channel).

If I wanted to play it only on headphones (which I sometimes do with my MiniDisc Walkman), I might choose to use the 3rd-party 'Crossfeed' DSP too.

And for MiniDisc I often use the 3rd-party 'Pause between tracks' DSP.

I usually get Case's FB2K installer which includes my favourite 3rd-party DSPs (and more) from http://www.saunalahti.fi/~cse/foobar2000 instead of the installer from the official Foobar site. This is less hassle that getting updated versions of the


If DSPs are used, typical DSP order is:

Resampler (SSRC)
<Convert mono to stereo>
<Pause between tracks>
<Crossfeed>
<I'd remove the Attenuator DSP (volume control) or make certain that it's at full volume = 0.0 dB>
Advanced Limiter

Advanced Limiter will look-ahead and catch any clipping peaks that remain (e.g. caused by resampling from a sampling rate other than 44.1 kHz) and deal with them with as little distortion as possible. In practice, with 89.0 dB ReplayGain volume, there's usually plenty of headroom to avoid clipping entirely, so Advanced Limiter will hardly ever interfere with the audio.

I believe this will give sound quality to virtually the maximum capabilities of the CD medium, given the source material, paying heed to all the "mathematically correct" principles of digital audio reproduction.

Regarding choice of CD-R media, for short-term use, most media are probably OK. For longer-term use/storage and less likelihood of burning a bad CD-R (or 'coaster'), the folks in the CD Hardware/Software forum tend to rave about Taiyo-Yuden and Mitsui CD-Rs (these are the manufacturers, and most brands periodically change supplier!). There's a forum thread on CDRlabs.com including pictures to help identify them. Taiyo-Yuden tend to be cheaper than Mitsui.

[Edit: typos]

This post has been edited by DickD: May 15 2003, 15:13

[2Bdecided]

I agree with almost everything that's been said, except...

If you're DJing, and intend to use pitch-shift dramatically, you do not want to use noise-shaped dither. It pushes the dither noise into high frequency (inaudible) regions. Selecting the right (or wrong ) pitch change on your DJ CD player could move this dither down an octave, right into one of the most sensitive regions of the human ear. This is the last thing you want to do.


Without noise shaping, the dither is spread evenly throughout the audio spectrum. With noise shaping, there's much less dither noise at audible frequencies, but a relatively large bump at high (inaudible) frequencies. If there's any chance at all of this bump being brought back down into the audible range, you should not use noise-shaping.


btw, I think the rules of this forum (which I sometimes don't adhear to myself) say that all reports of "I think X sounds better than Y" should be accompanied with a blind listening test proving that you actually hear a difference at all. This comment is directed especially to sxz, regarding his/her praise of the Wavelab 4 dither when burning mp3s to CD-R. If you do clearly hear a difference in a blind test, can you try comparing

1) mp3 decode truncated to 16-bits
2) mp3 decode dithered without noise shaping to 16-bits (1-bit tri dither)
3) mp3 decode dithered using your favourite dither
4) mp3 decode, dithered using your favourite dither and then re-dithered without noise shaping to 16-bits (1-bit tri dither)

...and let us know the results

Cheers,
David.

[d2e]

DickD: No matter how long the post, the info was well appreciated and offered more insight into the decoding process for my use.

2Bdecided: Your insight was most appreciated also. I will use pitch in mixing (don't know how dramatic it may be though) and may also find myself using a feature on some CD players known as "Master Tempo." Don't know if this would have any bearing on your suggestion. (Master Tempo = Locking in a song's pitch even when the tempo changes. Thus, I can vary the beat with no change in vocal or instrument tone.)

I'm going to try Dick's suggestion but take 2B's suggestion in doing "no noise shaping."

[Liquid_Predator]

How will the quality be if you decode with lame.exe --decode option? Can you als add dithering?

[Guest_Dex4now_*]

Hi d2e; I've read through this post twice, and I believe there might be an
important consideration thats being missed here. In your first post, you
mention doing DJ style mixing. Then, as the thread progresses, it appears as
though you may have been dissuaded from converting your mp3's to audio,
or .wav type discs.

I couldn't help but wonder if you plan on using a "pro-style" DJ mixing console
with "pro" CD players. I just did a quick perusal of MusiciansFriend and none of
their DJ style consoles play mp3's. (Or wave files, for that matter.)

If I've missed the boat here, I've only wasted a moment or two of your time.
On the other hand, I'ld hate to see you spend hours creating dozens
of discs that won't play on your intended equipment.

Dex

Edit: Ooops, left out an important point. Just, that depending on the
playback equipment you use, you may need to burn to .cda format.

This post has been edited by Dex4now: Jun 17 2003, 00:23