Hydrogenaudio Forums > mppenc 1.05g available (test release)

CiTay

Jun 7 2002, 15:27

Changes from 1.05f:

CODE

New option: --tag key=value (add APE tag entry using a string)

New option: --tagfile key (prompt)

New option: --tagfile key=filename (add APE tag entry using a file)

--ms 2 => --ms 3 (normal stereo which is used by --radio)

--ms 1 => --ms 2 (slighly reduced imaging)

--ms 1 new (heavily reduced imaging)

--ms 0 independent stereo

Small bugfixes not affecting quality

ATH changed again, more sensitive between 10 and 15 kHz now.

Adjusted low and high quality profiles. They are now equi-distant.

NMT changes from profile to profile by 1.5 dB.

--- Downloads removed: New test version available ---

*** This is for testing only, do not archive your music with it. ***

CiTay

Jun 7 2002, 15:40

Profiles for 1.05g:

CODE

--thumb OR --quality 3.0

NMT TMN MaxBand

3.5 11 19

--radio OR --quality 4.0

NMT TMN MaxBand

5 14 22

--standard OR --quality 5.0

NMT TMN MaxBand

6.5 17 28

--xtreme OR --quality 6.0

NMT TMN MaxBand

8 20 31

--insane OR --quality 7.0

NMT TMN MaxBand

9.5 23 31

Speek

Jun 7 2002, 18:49

The output of MPC 1.05g encoder is EXACTLY the same as the output of MP+ 1.7.9c! Apart from the bitrate (filesize) that is. I tested 4 files, charlies, dogies, newkid and thewayitis, with Eaqual and the output of Eaqual was the same every time. And I don't mean only the ODG, but every single value of Eaquals output matched exactly! How's that possible?

Bitrates---1.7.9c-1.05g
charlies---181---176
dogies----171---164
newkid----185---168
thewayitis-177---168

CiTay

Jun 7 2002, 19:09

I don't know what EAQUAL measures here. The output is far from being identical. Look in spectral view if you have doubts...

edit: what does it say about 1.01/1.02 and 1.04 output?

Speek

Jun 7 2002, 20:22

I got the files mixed up. Sorry. I'ld better get some sleep...

CiTay

Jun 8 2002, 16:45

Hehe Speek... shit happens.

Now that we got the EAQUAL stuff sorted out, i did an ABX test with 2nd_vent_clip.wav again, 1.05g --standard: 14/16 this time.

I think 1.05g is an improvement over 1.05f on this sample. The lowered-pitch-effect seems to be reduced for the first two HF sounds. I still can hear it quite reliably if i concentrate on the third HF sound though. It's hard to describe, so let me try to visualize it... on the pic, the first ellipse marks the first two sounds, the second ellipse marks the third sound.

removed

Maybe Dibrom and JohnV can also do another session... :you:

Frank Klemm

Jun 8 2002, 17:08

QUOTE

Originally posted by CiTay
Hehe Speek... shit happens.

Now that we got the EAQUAL stuff sorted out, i did an ABX test with 2nd_vent_clip.wav again, 1.05g --standard: 14/16 this time.

I think 1.05g is an improvement over 1.05f on this sample. The lowered-pitch-effect seems to be reduced for the first two HF sounds. I still can hear it quite reliably if i concentrate on the third HF sound though. It's hard to describe, so let me try to visualize it... on the pic, the first ellipse marks the first two sounds, the second ellipse marks the third sound.

removed

Maybe Dibrom and JohnV can also do another session... :you:

Try --ltq 16 or --ltq 17. These are 3 and 6 dB more sensitive in this
frequency range.

Frank Klemm

Jun 8 2002, 17:11

QUOTE

Originally posted by CiTay
Hehe Speek... shit happens.

Now that we got the EAQUAL stuff sorted out, i did an ABX test with 2nd_vent_clip.wav again, 1.05g --standard: 14/16 this time.

I think 1.05g is an improvement over 1.05f on this sample. The lowered-pitch-effect seems to be reduced for the first two HF sounds. I still can hear it quite reliably if i concentrate on the third HF sound though. It's hard to describe, so let me try to visualize it... on the pic, the first ellipse marks the first two sounds, the second ellipse marks the third sound.

rem

Maybe Dibrom and JohnV can also do another session... :you:

The pink spikes are showing that this sample is heavily clipped.
Reduce level of the original sample by 3 dB and test again.

Another question:

Who has a Linuxsystem. I have a ATH program to measure the ATH.
It needs Linux + a OSS supported soundcard.

CiTay

Jun 8 2002, 17:33

QUOTE

Originally posted by Frank Klemm

The pink spikes are showing that this sample is heavily clipped.
Reduce level of the original sample by 3 dB and test again.

Statistics for this sample:

CODE

Left Right

Min Sample Value: -32512 -32512

Max Sample Value: 32511 32511

Peak Amplitude: -.07 dB -.07 dB

Possibly Clipped: 0 0

DC Offset: 0 .033

Minimum RMS Power: -14.81 dB -14.05 dB

Maximum RMS Power: -4.74 dB -4.89 dB

Average RMS Power: -9.06 dB -8.85 dB

Total RMS Power: -8.92 dB -8.72 dB

(0 dB = FS sine wave)

Maybe non-full scale clipping..?

Frank Klemm

Jun 8 2002, 17:41

QUOTE

Originally posted by CiTay

Statistics for this sample:

CODE

Left Right

Min Sample Value: -32512 -32512

Max Sample Value: 32511 32511

Peak Amplitude: -.07 dB -.07 dB

Possibly Clipped: 0 0

DC Offset: 0 .033

Minimum RMS Power: -17.82 dB -17.06 dB

Maximum RMS Power: -7.75 dB -7.9 dB

Average RMS Power: -12.07 dB -11.86 dB

Total RMS Power: -11.93 dB -11.73 dB

Maybe non-full scale clipping..?

Yes. Very likely at -0.07 dB. Some soundcards begin to
distore -6 dB below FS, my Soundcard at -2 dB.

http://www.uni-jena.de/~pfk/mpc/audio2/overloadtest.zip

you should not hear the differential tone at 3.7 kHz.
The clip is 24 sec long and goes from -12 dB to +/-0 dB.

silver_cpu

Jun 8 2002, 17:56

QUOTE

--xtreme OR --quality 6.0

NMT TMN MaxBand
8 20 31

Is this to sa 31khz bandwidth? If so...is that really necessary? Maybe I'm just wrong, and it means something completely different.

Btw, am I correct in assuming that his test encoder still uses SV7?

And I have to ask while I'm at it...Frank, since you're planning on supporting 96khz input files, will you also be planning on support for 24bit input files? It'd be really great to just record live from my pro audio card, or be able to encode stereo DVD-A rips.

rjamorim

Jun 8 2002, 22:06

QUOTE

Originally posted by silver_cpu

And I have to ask while I'm at it...Frank, since you're planning on supporting 96khz input files

Where did you read that??

http://www.hydrogenaudio.org/forums/showth...10257#post10257

Case

Jun 8 2002, 22:11

QUOTE

Originally posted by rjamorim
Where did you read that??

MPC evolves. All samplerates will be supported.
http://www.uni-jena.de/~pfk/mpp/sv8/samplefreq.html

Garf

Jun 9 2002, 00:02

QUOTE

Originally posted by Frank Klemm

Who has a Linuxsystem. I have a ATH program to measure the ATH.
It needs Linux + a OSS supported soundcard.

Count me in.

--
GCP

silver_cpu

Jun 9 2002, 01:24

rjamorim: I read it on the official MPC website maintained by Klemm. Take a look at the "Latest" section on his site, and you'll see that it's on his "to do" list.

rjamorim

Jun 9 2002, 01:50

I see.

Thanks for the info.

Frank Klemm

Jun 9 2002, 02:07

QUOTE

Originally posted by silver_cpu

Is this to sa 31khz bandwidth? If so...is that really necessary? Maybe I'm just wrong, and it means something completely different.

Btw, am I correct in assuming that his test encoder still uses SV7?

And I have to ask while I'm at it...Frank, since you're planning on supporting 96khz input files, will you also be planning on support for 24bit input files? It'd be really great to just record live from my pro audio card, or be able to encode stereo DVD-A rips.

Is answered this question but a Netscape Windows resize discarded the text.
Thanks Dibrom for providing this nice interface.

Data transmitted since 2002-06-01:

209.197.243.159 28 MByte 61%
193.99.144.71 6 MByte 14%
Rest 11 MByte 23%
eMail <1 MByte 2%

Via my 56K modem this sucks.

rjamorim

Jun 9 2002, 02:09

QUOTE

Originally posted by Frank Klemm

Thanks Dibrom for providing this nice interface.

LOL

Frank Klemm

Jun 9 2002, 02:13

QUOTE

Originally posted by Garf

Count me in.

--
GCP

I set you on the mailing list.

CiTay

Jun 9 2002, 17:02

QUOTE

Originally posted by Frank Klemm
Some soundcards begin to
distore -6 dB below FS, my Soundcard at -2 dB.

Mine at -3 dB.

Okay, i tried reducing the level by 3 dB, but that wasn't it.

1.05g --standard --ltq 16, with the -3dB version: 14/16. First two tries wrong. But as i listened to the sample for the x-th time, i noticed another thing. The third HF "burst" diminishes slightly earlier in the encoded version. The lowered pitch-effect is almost completely gone now, the shorter HF stuff is the only thing i can ABX this with.

1.05g --standard --ltq 17: 16/16. Same as above. Lowered pitch almost gone, but third HF blow ends earlier. I think i heard this all the time, but never concentrated on it. Maybe it just sounds different now, but it doesn't sound so much like a lowered pitch anymore...