Hypercube Transcoder 2.0 beta is here !

[hypercube]

Hypercube Transcoder is a MPEG Multichannel encoder.

what's new ?:

Multichannel WAV as input
Batch encoding
Dynamic range processing
Normalisation

*BETA VERSION*: not fully tested. :D

http://hypercube.is.dreaming.org

[symonjfox]

I tried it and I find it a nice program!

There are 3 things I must ask:
1- A bitrate of 368 kbs, isn't it a bit high for a SVCD? I always use 112, 128 kbs giving more quality to the video (since the max total bitrate is small). Why isn't the bitrate modifiable? I think that about 60 kbs for each channel is a little high for SVCD use. Maybe a 56 or 48 kbs for channel should be ok.
2- Isn't it possible to create a 4 channel MPEG audio (instead of 6)? maybe it should pull down the bitrate, giving the sourround sound too.
3- Isn't it a way to create a Multichannel as Joint Stereo creates a 2 channel file? It should improve global quality, just writing down the main sound, and writing in the sourround channels, just the differences between those channels and the main. Is it possible? (Sorry for my english)

[hypercube]

first, let me tell you that I like your photo :D

>1- A bitrate of 368 kbs, isn't it a bit high for a SVCD?
not for 5.1 configuration. even at 384kbps, rear channels
are not very good (during splash symbal for example)

>2 & 3
encoding parameters will be accessible in future version.
Join Stereo will be possible and lower bitrate also.
but this is not for tomorrow, I am actually working on
another program to restore video tapes.

[symonjfox]

Ok, thanks for the replies :)

The photo is not mine ... he is Diego Abatantuono, an Italian comic actor, its movies are very dementials :D

[S_O]

I found a bug: 44100Hz 6ch wav input will be read like 48000Hz, the output mp2 is plays faster than the input and is shorter. That should be fixed.
Do you use your own encoder or do you simply copied the iso-dist10-source? Is your encoder more optimized than dist10 is?

[Cidici]

Originally posted by symonjfox

2- Isn't it possible to create a 4 channel MPEG audio (instead of 6)? maybe it should pull down the bitrate, giving the sourround sound too.


I think that creating 2-channel stereo with -s surround2 option in besweet could achieve a good surround quality using less bitrate.


Ps A come Atrocità, doppiaT come Terremoto e Traggedia, I come Iradiddio, L come Lagodisangue e A come Adesso vengo e ti sfascio le corna :D

[symonjfox]

Yes, I always use Sourround 2 for my SVCDs. I just wanted to know if a 4 channel would be possible ...

PS W ATTILA! Troppo bello quel film :)

[hypercube]

>The photo is not mine ...
I d'ont known Diego Abatantuono but he seems to be completely crazy.
I like that.
;)

>I found a bug: 44100Hz 6ch wav input will be read like 48000Hz
yes, this is not a bug. simply I don't compute upsampling.
I mean: 44100 => 48000.
I have forgotten messagebox saying this is impossible for instance.

check 44100 in output freq to avoid this problem.

>Do you use your own encoder or do you simply copied the iso-dist10->source?
:eek: AARRRRGl ! "simply copied" ! arf arf,arf !
yes, sure. simply copied. simply compiled, simply debugged
and simply ported to win32. All in one or two hours.

:p

>I just wanted to know if a 4 channel would be possible
in next version. sorry.

[S_O]

yes, this is not a bug. simply I don't compute upsampling.
I mean: 44100 => 48000.
I have forgotten messagebox saying this is impossible for instance.
check 44100 in output freq to avoid this problem.I tested with 48kHz and 44,1kHz. When I use 48kHz The file 55 sec instead of 1 minute, and output mp2 is 48kHz. When I select 44,1kHz it is also just 55 sec, but 44100kHz. The file gets resampled to 44,1kHz, as if it would be a 48kHz source.
AARRRRGl ! "simply copied" ! arf arf,arf ! yes, sure. simply copied. simply compiled, simply debugged
and simply ported to win32. All in one or two hours.
I guess it is simple compared to wite a new encoder from scratch.
I wanted to known, if
a) your encoding engine is identicall to dist10 (with same settings output is identical)
b) your encoding engine is based on dist10, you improved the encoder.
c) you wrote a completly new encoder or it´s based on a other encoder.
dist10 is anything else than optimized, it´s just an an example code. I don´t how your programming knowledge is and how much time you have, but MusePack (the highest quality lossy codec) is based on mp2 and Frank Klemm (a developer of MusePack) said once it shouldn´t be very difficult to output mp2 again with the encoder (quality will be little bit worse and bitrate a bit higher).
If you have enough time you could maybe use that the source of MusePack to create a high-end MP2-encoder.

[hypercube]

I tested with 48kHz and 44,1kHz. When I use 48kHz The file 55 sec instead of 1 minute, and output mp2 is 48kHz. When I select 44,1kHz it is also just 55 sec, but 44100kHz. The file gets resampled to 44,1kHz, as if it would be a 48kHz source.
if you select 44100 as output frequency, this is work, isn't it ?

a) your encoding engine is identicall to dist10 (with same settings output is identical)
yes

your encoding engine is based on dist10, you improved the encoder
no

you wrote a completly new encoder or it´s based on a other encoder.
no

dist10 is anything else than optimized, it´s just an an example code
yes, and buggy, and don't tell me how memory allocation is managed !
but it works. So I decided to plug azid as input to avoid the use of
theses #!@& multichannel AIFF files...
My own code is: resampling, normalisation and dynamic range processing.

said once it shouldn´t be very difficult to output mp2 again with the encoder
I think he was talking about stereo mp2, not Multichannel.
there is also tooLame (not lame !).

I don´t how your programming knowledge is and how much time you have,
I don't have so much time to do everything I want.:scared:
my programming knowledge is good but this is not suffisent to
build an MPEG encoder. I mean, there is a lot of mathematics in DSP.
:(

[hypercube]

hello DSPguru, you didn't respond to my last question about
NTSC / PAL audio convertion...
http://forum.doom9.org/showthread.php?s=&postid=252785#post252785

it seems you know every DSP programmers around here... :p

[S_O]

You can use soundtouch (open-source library): http://www.sunpoint.net/~oparviai/soundtouch/
You can convert the 24fps film for example to 25fps PAL
by setting tempo to 104,16% of the original tempo (25/24 * 100).
The sound pitch is not affected by this. This method is called "PAL-speed-up" and normally always done for Film -> PAL conversation (a other method is to copy the last frame of one second, frame 24 becomes frame 24 and 25).

[hypercube]

hypercube transcoder use azid API.
I tried to use liba52 on Win32
with all sort of AC3 and my opinion
is that azid is more robust.