Transcoding test

2007-04-13 13:50:08

So, someone will probably shoot me for this!

This little test was just done out of curiosity, for fun. It bares little relevance for real life situations.

Method:
1. I ripped a nice Yello track to wave. (Yello - Nervous)

2. I set up EAC with the following encoders:
[blockquote]AAC: iTunesEncoder, with iTunes 7.1.1.5 installed, set at 160 kbps VBR
Mp3: Lame 3.98 a11, V3 --vbr-new (~175 kbps)
Ogg Vorbis: aoTuV b5, Q5 (~160 kbps)[/blockquote]

I chose fairly high bitrates to sort of simulate a fairly conservative transcoding.

3. I did the following transcodings:
[blockquote]wave Original ->
AAC -> wave ->
Mp3 -> wave ->
Vorbis -> wave[/blockquote]

The "vorbis wave" then went through the whole process again, until I had a final wave that had been through each encoder five times.

The conversion to wave was done using the Wave filter in BonkEnc 1.0.1.

Results:
I haven't had time to do any ABX tests, and I don't really feel like doing any either  But here are the results so you can listen to them for yourself if you like:

Zip containing all of the following:
Flac sample of original wave
Flac sample of wave after 1 round of transcoding
Flac sample of wave after 2 rounds of transcoding
Flac sample of wave after 3 rounds of transcoding
Flac sample of wave after 4 rounds of transcoding
Flac sample of wave after 5 rounds of transcoding

I'd say the sample seems to deteriorate in a pretty linear fashion with each round of transcoding. After 1 round, I'd say it sounds better than I thought, but there seems to be less crispness in the high frequencies. For each round the crispness is reduced, and after five rounds of torture the sample is hissy and woshy like a 128 kbps blade

Transcoding test

Reply #1 – 2007-04-13 14:22:30

Here's an MPEG test on AAC tandem coding at 128kbps:


And the same test done with SLS at 512kbps: