m4a Conversion Adding Short Gap

[IrishPub]

I've recently started converting files for my iPod to m4a (rather than MP3 I was previously using). The source files are all FLAC. I've noticed that the conversion using the Nero AAC encoder adds a brief gap (0.0362 seconds to be precise) to the start of every file. As far as I can tell, no data is truncated, just the silence is prefixed. I only noticed this because some of the files are live recordings, and I could hear the gap when tracks change. Screen grabs from the files in Audacity are below.

I've also tried converting the FLAC files to intermediate WAV files, but when the WAV files are converted to m4a, the same gap gets added. (The gap is not present in the intermediate WAV.)

I can confirm that converting the files from FLAC to MP3 or from WAV to MP3 does not add this gap.

Is it normal that the Nero AAC encoder does this? How do I stop it?

[Spoon]

Re: m4a Conversion Adding Short Gap

Actually all lossy encoders have the gap (mp3 + aac), it is just for mp3 that decoders are mature enough to detect an remove the gap...

[IrishPub]

Re: m4a Conversion Adding Short Gap

Does that mean if I want gapless playback on my iPod, I have to use mp3 and not m4a/aac?

[garym]

Re: m4a Conversion Adding Short Gap

Does that mean if I want gapless playback on my iPod, I have to use mp3 and not m4a/aac?

Have you loaded up these m4a files into itunes? When files are added to the itunes library, it typically does a "gapless analysis" and then on playback on the ipod, does do gapless. But I use mostly mp3 files (LAME -V2) on my iThings so can't really test on my end.

[IrishPub]

Re: m4a Conversion Adding Short Gap

The way I discovered the m4a gaps was by listening on my iPod with the songs transferred via iTunes.

I just added some more m4a files into iTunes, and I don't see "gapless analysis" happening. I'm using iTunes 11, and importing of files is practically instantaneous.

Is there some way to force iTunes to do gapless analysis for m4a? Is there some way to see the gapless analysis data for each song? I can't find any settings like that.

[garym]

Re: m4a Conversion Adding Short Gap

The way I discovered the m4a gaps was by listening on my iPod with the songs transferred via iTunes.

I just added some more m4a files into iTunes, and I don't see "gapless analysis" happening. I'm using iTunes 11, and importing of files is practically instantaneous.

Is there some way to force iTunes to do gapless analysis for m4a? Is there some way to see the gapless analysis data for each song? I can't find any settings like that.

Not sure. Seems to do it automatically with mp3 files. It uses the info in the LAME header re: gapless as far as I know. But it is supposed to be automatic on m4a as well. Probably need to ask on an itunes/ipod forum. For example:

http://www.ilounge.com/index.php/articles/comments/gapless-playback-in-itunes-11/

or just return to mp3 for use on ipods. Testing demonstrates that at reasonably high bitrates, there is no quality to be gained in using m4a vs mp3. See hydrogenaudio.org for more info on such testing.

[IrishPub]

Re: m4a Conversion Adding Short Gap

Thanks. I'll have to check on some iTunes forums.

I've done extensive testing of my own on lossy files using the ABX feature in Foobar. The bitrate at which I cannot reliably distinguish in true blind tests between an original lossless file and a lossy version is 320kbps for mp3, and 256kbps for m4a/aac. The whole reason I was switching to m4a was to fit more tunes onto my iPod. (The objective wasn't to gain quality but to reduce file size for a comparable quality.)

[garym]

Re: m4a Conversion Adding Short Gap

got it. And using the ABX function in foobar2000 is the best way to determine, as you've already done. p.s. Wow, it takes 320kbs to reach transparency. You must either have golden ears (or be under 30 and yet to suffer normal age related high frequency loss). Or you've trained yourself to detect lossy coding artifacts. Anyhow, impressive. My own testing indicates I reach transparency in almost all normal music at mp3 192kbs. That's why I use LAME -V2. I've not bothered to test m4a. But with my 220,000+ library, I only include a random set of tunes on my iThings. And in my home system I'm using FLAC anyhow....

[IrishPub]

Re: m4a Conversion Adding Short Gap

I can't necessarily distinguish lower than 320kbps mp3 files in all cases, but there definitely are some files that I can. I had to pay really close attention doing the ABX tests, listening over and over and over with high end headphones. One of the best samples I found for distinguishing was a single clip just five seconds long from the Steely Dan song "What A Shame About Me" on the Two Against Nature album.

Interestingly in my tests, with mp3 files the giveaway was usually in the treble, but with m4a files it was usually in the bass! I hadn't expected the bass to differ, and I wouldn't have believed it if I hadn't heard it with my own ears in a blind test.

The whole ABX testing process was a fascinating learning experience. When I was right on the edge of transparency, one moment I thought I heard differences, but the next time I played the exact same clip I didn't. It really gave me a sense of how the mind plays sonic tricks on itself.

[garym]

Re: m4a Conversion Adding Short Gap

Yep, everyone should have to do a good ABX test. Many golden eared audiophiles would be quite surprised (given their usual claims of "night and day" difference between any mp3 and the lossless original.