WebSound.ru Home



  Поиск:

Поиск по WebSound.Ru:
Поиск в Интернете:
Powered by




  Партнеры, реклама:




Audio watermarking
TrustedAudio.com



 

Encoders express testing (part 2):
MS Windows Media Audio (.WMA), Ogg Vorbis (.OGG) and Liquid Audio (.LQT)

Author: (c)Alex Y. Radzishevsky
, http://radzishevsky.info
23.Sep.2000

 

Please, don't forget to read the part 1 of this testing (especially, the important remark at the end).

Tested compression algorithms: MS Windows Media Audio (.WMA), Ogg Vorbis (.OGG) and Liquid Audio (.LQT) in two different modes: MPEG-2 AAC and Dolby AC-3. Some coders do not allow to code on certain bitrates, so sometimes instead of one of encoders will be used Lame Encoder v3.86 (MP3).

Method of testing: coding of an original .WAV-fragment by each of codecs, back decoding to .WAV and comparison resulting amplitude-frequency response of original and received .WAV-files. Because Ogg Vorbis codes only on VBR and does not allow to specify bitrate definitely, the selection of coding parameters for all codecs was carried out so that the compressed files were approximately of one size.

Used encoders: OggEnc v0.4 (Ogg Vorbis), Windows Media Audio V7 (in a complete set Windows Media Encoder), Lame Encoder v3.86, Liquid Audio v5.0 beta 24 (includes MPEG-2 AAC and Dolby AC-3 encoders).

Fragments taken for testing: the first fragment (tst2.wav, 7056 Kb, PCM WAV 44.1 KHz, 16 bit, stereo) has duration 40 seconds, spectral structure with mainly average frequencies; the second fragment (tst4.wav, 10058 Kb, PCM WAV 44.1 KHz, 16 bit, stereo), duration 1 minute very much sated spectrum, is especial on high frequencies (Jean Michel Jarre, Oxygene VII).

Remarks: in the tests instead of LQT MPEG-2 AAC we shall write for brevity simply LQT; instead of Ogg Vorbis - simply OGG; LQT AC-3 - as is, LQT AC-3; MP3, created with the help of Lame Encoder v3.86 - MP3. CBR means "constant bitrate", VBR means "variable bitrate". All given diagrams are for the left channel (may be in the different scales).

Test 1 (64 Kbps) 


Original file tst2.wav (40 sec, 7056 Kb, PCM WAV 44.1 KHz, 16 bit, stereo)
Results and encoding parameters of MP3 LQT (AAC) WMA
bitrate 64 Kbps 64 Kbps 64 Kbps
file size 320 Kb 330 Kb 327 Kb
size of decoded WAV 3529 Kb (22 Khz) 3520 Kb (22 Khz) 7069 Kb (44 Khz)

Resulting amplitude-frequency response:

Resulting amplitude-frequency response in the bigger scale:



Test 2 (64 Kbps) 


Original file tst4.wav (1 min, 10584 Kb, PCM WAV 44.1 KHz, 16 bit, stereo)
Results and encoding parameters of MP3 LQT (AAC) WMA
bitrate 64 Kbps 64 Kbps 64 Kbps
file size 480 Kb 493 Kb 496 Kb
size of decoded WAV 5292 Kb (22 Khz) 5291 Kb (22 Khz) 10584 Kb (44 Khz, exact concurrence to the original)

Resulting amplitude-frequency response:

Resulting amplitude-frequency response in the bigger scale:



Test 3 (96 Kpbs) 


Original file tst2.wav (40 sec, 7056 Kb, PCM WAV 44.1 KHz, 16 bit, stereo)
Results and encoding parameters of MP3 LQT (AAC) WMA
bitrate 96 Kbps 96 Kbps 96 Kbps
file size 480 Kb 493 Kb 489 Kb
size of decoded WAV 3529 Kb (22 Khz) 3520 Kb (22 Khz) 7069 Kb (44 Khz)

Resulting amplitude-frequency response:

Resulting amplitude-frequency response in the bigger scale:

Original file tst2.wav (40 sec, 7056 Kb, PCM WAV 44.1 KHz, 16 bit, stereo)
Results and encoding parameters of LQT (AC-3) LQT (AAC) WMA
bitrate 96 Kbps 96 Kbps 96 Kbps
file size 482 Kb 493 Kb 489 Kb
size of decoded WAV 7060 Kb (44 Khz) 3520 Kb (22 Khz) 7069 Kb (44 Khz)

Resulting amplitude-frequency response:

Resulting amplitude-frequency response in the bigger scale:



Test 4 (96 Kbps) 


Original file tst4.wav (1 min, 10584 Kb, PCM WAV 44.1 KHz, 16 bit, stereo)
Results and encoding parameters of MP3 LQT (AAC) WMA
bitrate 96 Kbps 96 Kbps 96 Kbps
file size 720 Kb 739 Kb 741 Kb
size of decoded WAV 5292 Kb (22 Khz) 10583 Kb (44 Khz) 10584 Kb (44 Khz, exact concurrence to the original)

Resulting amplitude-frequency response:

Original file tst4.wav (1 min, 10584 Kb, PCM WAV 44.1 KHz, 16 bit, stereo)
Results and encoding parameters of LQT (AC-3) LQT (AAC) WMA
bitrate 96 Kbps 96 Kbps 96 Kbps
file size 735 Kb 739 Kb 741 Kb
size of decoded WAV 10591 Kb (44 Khz) 10583 Kb (44 Khz) 10584 Kb (44 Khz, exact concurrence to the original)

Resulting amplitude-frequency response:



Test 5 (128 Kbps) 


Original file tst2.wav (40 sec, 7056 Kb, PCM WAV 44.1 KHz, 16 bit, stereo)
Results and encoding parameters of OGG LQT (AAC) Streaming WMA
bitrate Mode 2 (VBR near 128 Kbps) 128 Kbps Streaming 128 Kbps
file size 661 Kb 657 Kb 651 Kb
size of decoded WAV 7056 Kb (exact concurrence to the original) 7057 Kb 7061 Kb

Resulting amplitude-frequency response:

Resulting amplitude-frequency response in the bigger scale:

Original file tst2.wav (40 sec, 7056 Kb, PCM WAV 44.1 KHz, 16 bit, stereo)
Results and encoding parameters of LQT (AAC) Transparent LQT (AAC) Streaming LQT (AC-3)
bitrate 128 Kbps Tranparent 128 Kbps Streaming 128 Kbps
file size 618 Kb 657 Kb 638 Kb
size of decoded WAV 7057 Kb 7057 Kb 7058 Kb

Resulting amplitude-frequency response:

Resulting amplitude-frequency response in the bigger scale:



Test 6 (128 Kbps) 


Original file tst4.wav (1 min, 10584 Kb, PCM WAV 44.1 KHz, 16 bit, stereo)
Results and encoding parameters of OGG LQT (AAC) Streaming WMA
bitrate Mode 2 (VBR near 128 Kbps) 128 Kbps Streaming 128 Kbps
file size 984 Kb 985 Kb 986 Kb
size of decoded WAV 10584 Kb (exact concurrence to the original) 10583 Kb 10584 Kb (exact concurrence to the original)

Resulting amplitude-frequency response:

Resulting amplitude-frequency response in the bigger scale:

Original file tst4.wav (1 min, 10584 Kb, PCM WAV 44.1 KHz, 16 bit, stereo)
Results and encoding parameters of LQT (AAC) Transparent LQT (AAC) Streaming LQT (AC-3)
bitrate 128 Kbps Tranparent 128 Kbps Streaming 128 Kbps
file size 1211 Kb 985 Kb 1103 Kb
size of decoded WAV 10583 Kb 10583 Kb 10585 Kb

Resulting amplitude-frequency response:

Resulting amplitude-frequency response in the bigger scale:



Test 7 (192 Kbps) 


Original file tst2.wav (40 sec, 7056 Kb, PCM WAV 44.1 KHz, 16 bit, stereo)
Results and encoding parameters of OGG LQT (AAC) WMA
bitrate Mode 4 (VBR near 192 Kbps) 192 Kbps  192 Kbps
file size 887 Kb 834 Kb 975 Kb
size of decoded WAV 7056 Kb (exact concurrence to the original) 7057 Kb 7061 Kb

Resulting amplitude-frequency response:

Original file tst2.wav (40 sec, 7056 Kb, PCM WAV 44.1 KHz, 16 bit, stereo)
Results and encoding parameters of LQT (AC-3) LQT (AAC) WMA
bitrate 192 Kbps 192 Kbps  192 Kbps
file size 934 Kb 834 Kb 975 Kb
size of decoded WAV 7058 Kb (exact concurrence to the original) 7057 Kb 7061 Kb

Resulting amplitude-frequency response in the bigger scale:

Test 8 (192 Kbps) 
Original file tst4.wav (1 min, 10584 Kb, PCM WAV 44.1 KHz, 16 bit, stereo)
Results and encoding parameters of OGG LQT (AAC) WMA
bitrate Mode 4 (VBR near 192 Kbps) 192 Kbps Streaming 192 Kbps
file size 1501 Kb 1634 Kb 1476 Kb
size of decoded WAV 10584 Kb (exact concurrence to the original) 10583 Kb 10584 Kb (exact concurrence to the original)

Resulting amplitude-frequency response:

Resulting amplitude-frequency response in the bigger scale:

Original file tst4.wav (1 min, 10584 Kb, PCM WAV 44.1 KHz, 16 bit, stereo)
Results and encoding parameters of LQT (AC-3) LQT (AAC) WMA
bitrate 192 Kbps 192 Kbps Streaming 192 Kbps
file size 1546 Kb 1634 Kb 1476 Kb
size of decoded WAV 10585 Kb (exact concurrence to the original) 10583 Kb 10584 Kb (exact concurrence to the original)

Resulting amplitude-frequency response:

Resulting amplitude-frequency response in the bigger scale



Brief conclusions 


WMA has simply surprised on 64 Kbps: while all coders have cut off a signal on 10 Khz (giving 22-kilohertz output files), WMA has fairly fulfilled all frequencies up to 20 Khz, and done this besides qualitative enough for such low bitrate. But it is necessary to assume, that such achievement on all frequencies, probably, has an appreciable effect on the signal quality in the lower strip of frequencies, but also here impossible to say anything bad: during listening WMA sounded appreciably better than opponents. Thus, preview of songs, which are usually done using MP3 64 Kbps, would sound much better if they was coded using WMA. MP3 and LQT (AAC) are not remarkable - the give out files WAV on 22 Khz, which means, with frequencies limited by 10 Khz.

On 96 Kbps LQT (AAC) behaves obviously better than MP3: the strip of saved frequencies has extended up to 15 Khz, that, certainly, has a very appreciable effect on sound. However, nevertheless, WMA all the same is in the lead in quality. As to LQT Dolby AC-3, by the diagram he behaves slightly better, than LQT AAC. But as to subjective recognition - it is difficult to determine the leader between LQT AAC and LQT AC-3. 

In the test on 128 Kbps instead of MP3 participated OGG (because its minimum average bitrate is 128 Kbps). LQT AAC in this test has cut off all above 18 Khz, though its sounding is absolutely good. But here has lagged behind yet WMA - OGG obviously fulfils all strip of frequencies even not interrupting on 20 Khz . The sounding of LQT, WMA and OGG in this test begins difficult to compare, and it is necessary to believe the diagrams. But, nevertheless, it seemed, that WMA and OGG sound better than LQT, though it can be simply my preconceived opinion. As to LQT AC-3, he is simply ñ00l: its diagram well repeats amplitude-frequency response of an original signal. Its sounding is obviously very good too. In this test, by the way, participated LQT AAC Streaming and LQT AAC Transparent, any unexpectedness had not taken place - they behave equally.

On 192 Kbps there was happened precisely what I supposed right in the beginning: the diagram of WMA has lagged behind on accuracy from LQT and OGG. And there is one explanation. WMA even on 64 Kbps fulfilled almost all spectrum and it indirectly means that WMA uses an "artful" psychoacoustic model. And, now it proves to be true evidently on the diagram. LQT, though did not fulfil so qualitatively all the spectrum on the top, he nevertheless, probably, repeated an original signal more precisely in fulfilled range. WMA at the expense of some shifts with a spectrum showed in the whole best reaction to all frequencies, but, thus, the signal by it was deformed. OGG shows on 192 Kbps quite good results and, probably, it appears a little bit better, than WMA. LQT AC-3 - is very good, it's diagram precisely repeats the diagram of the original signal. Who is the best - OGG or LQT AC-3 - is difficultly to say.

In the whole. MP3, unfortunately, is noncompetitive. LQT AAC on the low bitrates saves the limited spectrum of frequencies, but thus, probably, deforms the signal a little. On high bitrates LQT AAC shows quite good results. WMA at the expense of use obviously artful psychoacoustic model shows on the low bitrates much more good results, than the opponents, on the high bitrates results remain quite good, but the distortions of a signal on the diagram can be noticed easily. It is very difficultly to say how (bad or good) it influences on perception of sounding. As a whole WMA shows one of the best results. OGG participated not in all tests owing to impossibility of coding on 64 and 96 Kbps. In the tests on 128 and 192 Kbps OGG has shown the best results. Who is the best - OGG or WMA is very complex decision, but the diagram of OGG nevertheless is exacter. As to LQT Dolby AC-3, it is very qualitative codec. And, if on low bitrates he does not show the best results, on 128 and, especially, 192 Kbps is a very qualitative decision.