Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Protocol of my 2+ hrs telephone conversation (Read 48816 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Protocol of my 2+ hrs telephone conversation

Reply #100 –
@Ivan Dimkovic

You have read the thread and know more than most of us about AAC. Could the benefits of MPC be applied to AAC as Frank stated? If so it would be amazing to see Nero hire him to optimize AAC for transparency (since that has been his priority and most AAC developers seem to have the priorities set on lower bit rates)

Protocol of my 2+ hrs telephone conversation

Reply #102 –
Hmm, maybe if all of us using MPC donate some money we can get SV8? I'd be willing to pay as much as $100 for SV8 with all original intended features and Matroska support.
The object of mankind lies in its highest individuals.
One must have chaos in oneself to be able to give birth to a dancing star.

Protocol of my 2+ hrs telephone conversation

Reply #103 –
It doesn't seem to me that money is what Frank needs to do SV8. It seems to be motivation.

Protocol of my 2+ hrs telephone conversation

Reply #104 –
... and maybe a little bit more time was nice also, according to him ....

Protocol of my 2+ hrs telephone conversation

Reply #105 –
Quote
It doesn't seem to me that money is what Frank needs to do SV8. It seems to be motivation.

Thank you!  From what I have understood during previous conversations with HA personnel close to Frank, he is very discouraged with the lack of participation and input from the HA community on test builds.  All he sees when he visits HA is "where's sv8?!" no one willing to help out with testing, just people eager for another release without putting in any of the work necessary to build up to that release.  And now with Frank's frustration now publicized on HA for everyone to see, everyone abandons MPC development and suggests that Frank move on to an already thriving format with more than apt developers behind it.  Why?!  People that frequent this forum need to be willing to get involved in codec development supplying whatever talents they may have.  HA is the cornerstone of audio codec development, there is no middle man, we directly converse with developers of the leading audio codecs available to the public today... which means that we ultimately have a say in what happens with our favorite codes and where they go.

Protocol of my 2+ hrs telephone conversation

Reply #106 –
@ProtectYaNeck
Huh?
help with what? how can people help when there isn't anything to HELP?

I've lots of respect for what Mr. Klemm has done for the community. But this SV8 thing is just handled insanely bad. A clear example of lack of information leading to messy stuff.
What about him showing some interest and maybe asking for specific stuff that needs helping?

Assuming that people will flock to some unknown quest (that they aren't even sure is alive... at all) is just ludicrus.

Still, sv7 is awsome and I personally don't care much about sv8. It would be great if it ever gets here, but realistically it is a stretch. If it does get here, I guess AAC would have already surpassed it technically and it already has done so with compatibility.

The only thing thats been nagging me is the lack of information. Why hasen't Mr. Klemm said something like
1) My interests have changed, sv7 will be last version.
2) I've no time, work is done on sv8, but in small sessions with months between. Don't expect anything in years time
3) I've no time, i'd appreciate help on x, y, z,

Or something similar. 1-2 lines of text, estimate time to write, 20-25 seconds. Time saved from reading whining/requesting/questions emails and forums, Unlimited. Plus sv8 might not be in a standstill as its been for ages now.

This is just my opinion and how I've percieved the situation over the years, it could all be BS.

Protocol of my 2+ hrs telephone conversation

Reply #107 –
Personally,

I think that the best thing that could happen to Musepac is for Frank to share his technology with someone else, either with an an open source project, or license it, to one of the major players in AAC.  The guy is a genius, but not a businessman.
 

Protocol of my 2+ hrs telephone conversation

Reply #108 –
I think I mentioned this somewhere in the forums over a year ago. The psymodel for MPC has been open sourced right? Ogg Vorbis is open sourced right? There's nothing stopping anyone from porting the psymodel from MPC to be used in Vorbis right? Hmm...

Protocol of my 2+ hrs telephone conversation

Reply #109 –
Quote
I think I mentioned this somewhere in the forums over a year ago. The psymodel for MPC has been open sourced right? Ogg Vorbis is open sourced right? There's nothing stopping anyone from porting the psymodel from MPC to be used in Vorbis right? Hmm...

Hrmmm I think your on to something there tangent...

Isn't the psymodel tuned for whatever codec its going to be used for? E.g. tweaking the psymodel to better suit a codec.

Don't qoute me on that though, but it seems logical that this would be the case considering all codecs have differing methods on how they compress audio.

Regards

AgentMil
-=MusePack... Living Audio Compression=-

Honda - The Power of Dreams

Protocol of my 2+ hrs telephone conversation

Reply #110 –
Quote
Quote
I think I mentioned this somewhere in the forums over a year ago. The psymodel for MPC has been open sourced right? Ogg Vorbis is open sourced right? There's nothing stopping anyone from porting the psymodel from MPC to be used in Vorbis right? Hmm...

Hrmmm I think your on to something there tangent...

Isn't the psymodel tuned for whatever codec its going to be used for? E.g. tweaking the psymodel to better suit a codec.
AgentMil

I always thought the same, but Frank told me recently that a psy model can well be ported. He also had plans to make his psy model kind of a 'plugin' , for this reason.

Both Vorbis and Wavpack4 could benefit form this, same is valid for FAAC of course ....

Protocol of my 2+ hrs telephone conversation

Reply #111 –
Quote
Quote
It doesn't seem to me that money is what Frank needs to do SV8. It seems to be motivation.

Thank you!  From what I have understood during previous conversations with HA personnel close to Frank, he is very discouraged with the lack of participation and input from the HA community on test builds.  All he sees when he visits HA is "where's sv8?!" no one willing to help out with testing, just people eager for another release without putting in any of the work necessary to build up to that release.

I cant agree more. Frank doesnt like webboards, so i set up several mailing lists since a long time on http://freelists.org, and was trying to motivate other developers to start serious technical conversation about MPC related features. There were some nice posts, but then the traffic died again.

Frank was subscribed over 3 months to the matroska mailing lists, before he sent a first reply email. I had hoped the same could happen with MPC, but it seems there is no way to motivate other developers to start to feel responsible about it. Its like everybody staring at Frank, like the mouse at the snake .......

About Frank's motivation : Well, in the past weeks he has designed a new subtitles standard, invented a video-HiFi system with a new structure based on built-in, device specific decoders instead of transferring video signals in analog wires ( he IS a genius, no doubt ), and many other things also .....

...... it appears to me that the little spare time he has, is all invested into DVD Video related stuff now  .....

.. sad, but true. I was hoping that his interest in MPC could be reactivated once we can use it in matroska for video compression, but the only one being able to make SV 7.5 seeems to be him. In short, MPC development is caught in a classical catch 22  ......

Protocol of my 2+ hrs telephone conversation

Reply #112 –
Well, if the MPC psymodel is open source, then lobby to get the vorbis developers to adopt it.  They are getting ready for a complete overhaul anyway.  From the posts that I have seen above, the psymodel is the important part.

Protocol of my 2+ hrs telephone conversation

Reply #113 –
hmm, I don't know exactly, what F. K. develops for DVD-Video, like subtitle standard etc,
imo the DVD-V is already a standard, so what needs there to be done ?
and there are a lot easy & quick backup solutions available for DVD-V.
Burning DVD-+R/RW is way easier these days than X/S/VCD or divX/mpeg4.

And in a short time: discs capable of 20 GB, blue ray disc and related.

I can only pray to him, that he finishes the mpc project, with SV 7.5, pleaaaaaaaase !
and that mpc can be implemented into video/audio container, matroska.

Protocol of my 2+ hrs telephone conversation

Reply #114 –
What would it take to just get decent seeking capabilty with SV7? That's really all I am missing with the format.
flac > schiit modi > schiit magni > hd650

Protocol of my 2+ hrs telephone conversation

Reply #115 –
Quote
What would it take to just get decent seeking capabilty with SV7? That's really all I am missing with the format.

What do you mean by seeking capability? Are you talking about how the audio stream can occasionally get screwed up (a short burst of very loud noise) if you seek a track in WinAMP? Annoys the heck out of me and various in_mpc plugins haven't fixed the problem.

Protocol of my 2+ hrs telephone conversation

Reply #116 –
No, I mean just moving the slider to a forward position in the file. Like jumping ahead to the 3:00 mark in a 4:00 min song.

That doesn't seem to work very well. Part of that might be due to the fact that I'm playing these from a network file server. Vorbis seeks immediately. No delay at all. MP3 is pretty close to Vorbis.

Musepack doesn't do this for me, even in Foobar2000. My understanding was that it has to re-read the entire file (up to the point you are seeking to).

It's not something I use very often. Usually only when I need to hear a certain part of a song repeatedly.
flac > schiit modi > schiit magni > hd650

Protocol of my 2+ hrs telephone conversation

Reply #117 –
Quote
Isn't the psymodel tuned for whatever codec its going to be used for? E.g. tweaking the psymodel to better suit a codec.
Don't qoute me on that though, but it seems logical that this would be the case considering all codecs have differing methods on how they compress audio.

Nope. Psymodel is the modeling of the physics of the ear and is independant of coding method. If you look at the process of encoding audio, from the input it forks into two process (maybe after the filterbanking). The first process is the psymodel and the second process does the encoding. Once the psymodel process is completed, the information is passed over to the encoding process which then decides from there how much bits to allocate to each band.

I've mentioned before that ultimately, a transform coder will be more efficient than subband coder even at high bitrates once enough work is done on it. It's not hard to see why, one of the advantages is that a transform coder is able to group the bands in accordance to the critical bands, while a subband coder has to stick to the 32 equi-bandwidth bands. MPC's advantage over other codecs came from two factors: it's advanced psymodel and that transform coders have not completely solved the kinks in the system (e.g. preecho handling).

Protocol of my 2+ hrs telephone conversation

Reply #118 –
Quote
Quote
What would it take to just get decent seeking capabilty with SV7? That's really all I am missing with the format.

What do you mean by seeking capability? Are you talking about how the audio stream can occasionally get screwed up (a short burst of very loud noise) if you seek a track in WinAMP? Annoys the heck out of me and various in_mpc plugins haven't fixed the problem.

I think he means that, to jump to a point in time, the whole mpc has to be decoded to the given point. I assume this will be fixed in the Matroska implementation, whether that ends up being a variant of sv7 or sv8, as they have to fix the  framing anyway.

Protocol of my 2+ hrs telephone conversation

Reply #119 –
Quote
No, I mean just moving the slider to a forward position in the file. Like jumping ahead to the 3:00 mark in a 4:00 min song.

...

Musepack doesn't do this for me, even in Foobar2000. My understanding was that it has to re-read the entire file (up to the point you are seeking to).

Strange. I find MPC seeking to be instantaneous in WinAMP but perhaps it reads the entire file to memory, which would allow lightning access. MPC files decode very quickly as well (like 75x real-time on my ancient 800MHz Celeron) so my guess is that your media player doesn't cache the file and network latency delays the seeking.

Protocol of my 2+ hrs telephone conversation

Reply #120 –
Quote
Both Vorbis and Wavpack4 could benefit form this, same is valid for FAAC of course ....


Sorry but I disagree here. Wavpack does not need a psy model based on that utilised in mpc, at least in my opinion anyway. Currently Wavpack lossy does not use a model to discard any bits. The model that has just been introduced to Wavpack 4 alpha 2 determines the noise/hiss that may be perceptible and uses more or less bits to suit, but no masking/filtering etc is utilised, which makes Wavpack lossy much more useful than the other lossy codecs for transcoding in my experience.

In my own transcoding tests, mpc was clearly better than Vorbis or LAME into ATRAC for MD, but Wavpack was signigicantly better again. (As is also Optifrog Dualstream for similar reasons.)

A Psymodel that applies filters in Wavpack lossy? No thanks. A model for determining the quality of an encoded file so that it can be improved, or bits can be saved is fine, but at least the encoder is still processing the entire signal, and not messing it up with pre-echo, post-echo and other annoying artifacts.

Vorbis? Well that may be a completely different story altogether.

Den.

Protocol of my 2+ hrs telephone conversation

Reply #121 –
Quote
Strange. I find MPC seeking to be instantaneous in WinAMP but perhaps it reads the entire file to memory, which would allow lightning access. MPC files decode very quickly as well (like 75x real-time on my ancient 800MHz Celeron) so my guess is that your media player doesn't cache the file and network latency delays the seeking.

Create a large MPC file (e.g. CD image), and immediately after opening (which should be instantaneous) try a seeking operation to the end of the file. There will be a noticeable lag.

It is a known issue with SV 7, but it is usually no problem for short files.

Protocol of my 2+ hrs telephone conversation

Reply #122 –
Quote
Quote
No, I mean just moving the slider to a forward position in the file. Like jumping ahead to the 3:00 mark in a 4:00 min song.

...

Musepack doesn't do this for me, even in Foobar2000. My understanding was that it has to re-read the entire file (up to the point you are seeking to).

Strange. I find MPC seeking to be instantaneous in WinAMP but perhaps it reads the entire file to memory, which would allow lightning access. MPC files decode very quickly as well (like 75x real-time on my ancient 800MHz Celeron) so my guess is that your media player doesn't cache the file and network latency delays the seeking.

Yes, I think it's a combination of all of those things. If I set the cache high enough that it reads the entire file into memory, then the seeking is much improved. That has the unfortunate downside of slowing down the load time/start time. I might be able to "tune" this somewhat.

Edit: It's a 100Mb network connection, however it seems to be more like 50Mbs when doing an FTP.
flac > schiit modi > schiit magni > hd650

Protocol of my 2+ hrs telephone conversation

Reply #123 –
Seeking problems are only at the start of the file... eg. I open a 4min long song but if i try to seek straight away there is a noticeable lag between me clicking and the song seeking to that position. After that the file seeks almost perfectly.


Regards

AgentMil
-=MusePack... Living Audio Compression=-

Honda - The Power of Dreams

Protocol of my 2+ hrs telephone conversation

Reply #124 –
As Indybrett pointed, Musepack (SV7) seeking requires entire file until destionation point to be read. Usually not annoying on modern machines while playing small files, but can get very nasty with CD images played from a network drive over LAN connection. Most of decoder implementations have some kind of run-time generated seektables, so seeking gets faster if given point of file was reached by decoding/seeking process before.
Microsoft Windows: We can't script here, this is bat country.