Obviously, it's name seems to imply that it can be used for transcoding. By this, I assume that it would be "safe" (or, at least, "safe enough") to use if I later used this file as the source file (as opposed to going back to the original CD) if I wanted to convert it to a lower bitrate MP4 file.

But, what about using it as a source file if I wanted to convert it to a lower bitrate MP3 file for use for burning a MP3 compilation disc, etc? My understanding, from what I learned through these excellent forums here at hydrogenaudio, is that transcoding from one lossy format to another lossy format (or even transcoding to the same lossy format but just to a lower bitrate) is a big NO-NO, regardless of the bitrate of the lossy file you are using as the source file when transcoding. The reason, obviously, is due to the fact that noise/artifacts will be increased and/or effect the quality significantly more than if you had went back to the original CD as the source, which makes perfect sense.

But is there something special about Nero's "for transcoding" profile that makes it "safe" or, at least, significantly "safer" than one would normally expect based on what we know about the whole way lossy encoding normally works? Is it doing something significantly different/special in the way it is handling noise/artifacts that makes it inherently safer than it would normally be as opposed to, let's say, you had just used a ridiculus CBR of 320 or something using "normal" algorithms?

I guess the best way to really test this would be to find a sample where I wouldn't be able to tell the difference between, let's say, LAME alt-preset standard and the original CD when encoded to MP3 right from the original CD but where I would be able to tell the difference between LAME aps and the original CD when encoded to MP4 ("for transcoding") from the original CD then trancoded again from MP4 to MP3 aps. Of course, finding such a sample is usually easier said than done. So if anyone knows of a sample which does a good job of illustrating more noticable quality loss during transcoding and/or knows something that the "for transcoding" profile does that makes it "special", I'd look forward to hearing about it.

On a side note, since listening test are pretty useless at these very high bitrates, I had resorted to actually "looking" at the sound (after converting back to WAVE from MP4 "for transcoding") using Nero's wave editor and comparing the spectrogram, wavelet, etc. (although I know that this is not a good way to "test" and does not accurately reflect the true "quality"). But I was quite surprised at how similar the graphs looked between the original WAVE and the MP4 "for transcoding", even for very high frequencies. In some cases it was downright scary and certainly "looked" much more accurate than anything else I've seen before. So maybe there is something special going on here with that profile? But, since you really can't go by how the graphs look, I'm not putting much weight in it at all.

The 'special' thing that's going on is that the sounds are not lowpassed until 22,000 Hz...like the original cd. Most lower bitrates and other codecs lowpass much lower to avoid certain high frequency artifacts.

All the transcoding preset really does is provide the extra bitrate required for this high frequency information. So you're less likely to get ringing artifacs re-encoding this file or encoding to this bitrate from another lossy format. Using spectrogram is not a good way to test... but it can be usefull to determine if an unknown cd-r was burned from an mp3 copy (look for the lowpass at 16khz ish)...or if you want to see visually what some very bad cases of ringing actually look like in the frequency domain. This isn't a whole lot different the cbr 320 in principal, aside from the codec being a little more intelligent in the # of bits it alocates for a given frame.

Otherwise it's not a very reliable method to compare files because the way the spectrogram is generated is in itself lossy (the resolution is going to be less than the wav) both in the frequency axis (which is infinite) and the time domain (unless you make the graph so there is a pixel for every sample, even this should still generate errors though)

I don't know about AAC transcoding capabilities, but other 2 codecs are recognized (some test have been done on this, use the search function to find them) as very good for transcoding purposes. One is MPC (i think it's because it has no frequency trasformation, works on the time domain), the other is Wavpack lossy (version 4 final to be released very soon).

I would normaly use MPC around 180 kbps (probably lower than Nero AAC transcoding) and Wavpack lossy around 360 kbps (probably higher than Nero AAC transcoding), more or less.

¿Is there any reason (like having an iPod) to use AAC?

Yup, you hit the nail on the head - iPod.

Basically, what I am doing now is ripping to FLAC, then converting from FLAC to MP3 (alt-preset standard or alt-preset extreme) for burning MP3 CDs to play in the car. Then, also converting a second time from FLAC to AAC for iPod since I can go a bit lower on the bitrate to save some room. Then I file the FLAC files away on my hard drive incase I ever want to encode to a different bitrate or format again in the future or loose my original CD, etc.

So, even if I ripped from CD to Nero's Transparent, I would still being doing the same steps (i.e. going to MP3 and also going to a lower bitrate AAC) so I wouldn't actually be saving any steps. But I would be saving quite a bit of disk space.

But, from the sounds of it, I think I'll just stick with FLAC. Or, if disk space becomes more of an issue, I'll look into MPC and Wavpack lossy as you recommended.

Thanks for the help.

My iPod component for Foobar2000, foo_pod, can transcode from any Foobar2000 supported format directly to MP3 (using LAME) or AAC (using FAAC or Nero AAC with nencode or NAACEnc). Since it transcodes directly to the iPod, you not only save a step, but you also save disk space.

Nero "Transcoding" profile was optimized for multiple generations encoding - i.e. signal to noise ratios in the psych model are lifted in such way so the content would survive another couple of encodings.

Nice, I didn´t know that

I know this is a very difficult question, but how does it compare to the "hybrid" audio formats? And to subband encoders (MPC)?

After all, even the highly-sophisticated modern lossy codecs are exactly what they are called - they are LOSSY.

In my opinion, you should use a lossless or hybrid codec if you are really up to transcoding. HDD space is cheap so you can keep lossless backups of your valuable CD's that can then be transcoded to any format (even the ones not even developped yet) you like.

Hint @ Ahead

Nero AAC Lossless would be nice ...