I think this is a bug

[PeteC]

I've sent this as an email to Support but I'm interested to see if anyone else can reproduce it.

I think I've found a major bug in version 4.1.

Conditions: Windows XP Pro with SP1. AMD Athlon 2200+, 512MB DDR RAM. Leadtek A180 AGP graphics card. Monitor in 1280x1024x32bit mode.

All slides are jpg, 1024x768x24bit. I am making a menu slide by placing transparent buttons as objects onto the base slide, then setting their actions as Go to slide number...

1. Start a new project. Using any image as slide number 1 for a background, go into Object Editor.

2. Make a Button. Accept the default width and height, and set its properties to 'no button text' (just erase the word Button), Background Transparent. Position it anywhere over the background. (It's going to be a clickable hot-spot to jump to a slide further down the list.)

3. Click Copy, then Paste, and position the copy away from the first button. Up to four copies are OK, but do this five times and then click OK to exit the Object Editor.

An error message appears - "Access violation at address 00403864 in module 'apr.exe'. Read of address 00000017".

The address numbers vary depending on which object on the screen is clicked. If the OK button is clicked, the addresses will be all zeros, for example.

There is no way out of this except by using Task manager to end the program. This is absolutely reproduceable for me.

I have reverted to using v4.01 and do not get this problem with that version. I have noticed that in v4.1, I can finely position the buttons using the arrow keys, whereas in v4.01, I can't. Maybe it's to do with this new positioning feature.

Can anyone else reproduce this?

Regards,

Peter Croft

[Hawk]

Peter

I have a demo menu show which I created and encountered the exact same problem you have mentioned, same errors.

I have been in contact with Igor about this, and he could not reproduce it.

I have posted my complete show for him to download but have not heard back.

I understand he may have gone on vacation for a short period, we may learn more when he returns.

v4.11

Ralph

[think(box)]

Talk about a freaky coincidence, I just reported the same bug to Igor by email before reading this post!

I gave him a step-by-step duplication and .PTE file, but now I see that possibly any project will fail similarly.

FYI, all I had to do was trying to change the spelling of a linked .exe file in a button action. The properties sheet "OK" worked without trouble, but the fatal bug happened when I tried to OK the visual object editor window.

[PeteC]

Hmmm, thanks guys. Glad to know I'm not imagining it.

Oh well, I have plenty more work to do before I need that menu slide, so I'm not in a rush.

Cheers,

Pete

[Guest guru]

Thanks Peter, Ralph and Bill. Reproduced the same test, with identical issue.

Now I ask you to make another test.

Michel wrote me signaling an abnormal behaviour of synchronized shows using .ogg background files. I made a little test, and I think there is another bug in 4.10-4.11.

Please make a little simple show with 10-15 random pictures, add a little .ogg music file, place automatically transition points (without transition effects is better) and synchronize exactly one slide or two (any slide) with a beat of music.

Make .exe and save .pte file.

Now reopen .pte file, and verify the synchronization in the time line, starting not only by the beginning but also by any point before the synchronized slide...

And then please tell me something about the results. Thanks!

[think(box)]

I don't have any .ogg sound files, so unless you email one to me or post a link for download, I can't do the .ogg timing tests.

[Ken Cox]

that free download from

http://www.dbpoweramp.com/dbpoweramp.htm

will convert your files yto several formats that i can hardly pronounce

powerAMP Music Converter (dMC), is one of those must have utilities, indispensable for converting audio files from one format to another. Part of dMCs formidable arsenal, is its ability to rip music straight from Audio CDs!, or LPs [optional install].

Straight out of the box dMC can convert to and from Wave, Mp3 and from CD Audio (CDA) files.

Codec downloads add Windows Media Audio v8 (WMA) , Ogg Vorbis and many more types to the table. Features include:

Volume Normalization - have all audio tracks use same volume.

ID Tag preservation & Editing - keep those ID tags between conversions,

Explorer Audio Info Pop-ups - display useful information on an audio file in explorer,

Convert To - simplest way of converting right click on a file and select Convert To,

Send audio to your Portable Mp3 Player with Sveta Portable Audio

Fully compatible with Windows 95/98/ME NT4/2000 and Windows XP + Linux (when using Wine).

Round of a perfect package. What's more dMC is free. !! News Just in dMC is the Fastest Audio CD Ripper !!

dBpowerAMP software does not contain any 'Spyware' 'Add sponsorship' or other nasties

ken

[Guest guru]

Actually dBPowerAMP is an excellent and very useful software (but the most interesting is the converter: http://www.dbpoweramp.com/dmc.htm).

Anyhow, if someome want to get a little .OGG file "ready to use", I can willingly send it (1 min. = 300-400 KB).

[JRR]

and it is free too !!!!!!!!!!!!!!(dBPowerAMP)

I discovered it a few years ago when I was working with my Arion Wizard slide show firmware with my 35mm slides

Remember those funny little square things called slides ?? (I still do most of my shows in both 35mm and digital)

[think(box)]

Thanks for the tips! I downloaded and installed dBpowerAMP and a boatload of add-on stuff. It feels like I just installed Photoshop

Nice player! How do I make the player repeat a track without a playlist?

I am doing a conversion to Ogg Vorbis as I type.

[Ken Cox]

dont ask me -- i just use it to convert stuff

i can play the oog files with irfanview and with Media Jukebox so far

can hear no difference between mp3 and oog -- but then i had my tin hat on

but i opened the original file and the converted in soundforge 6 and the sine wave patterns appear the same

ken

[Guest guru]

can hear no difference between mp3 and ogg

But what is the size of files, Ken? I think .OGG is smaller!

[Ken Cox]

Guido

ran another conversion

original file

159 kb/128 bit mp3

to

oog file

158 kb/128 bit oog

btw Windows Media player 9 will not play and checked the WMP site for plugins -- no luck

Audacity crashes when i try it

Irfanview, Sonic Foundry and Media Jukebox and dBpowerAMP play it

ken

[Guest guru]

Dear Ken,

1. Please forget WMP... especially v9 !

2. When you compress a sound file, if possible start always by a .wav file, not by an already compressed format like mp3.

3. Try lower bitrates converting to OGG format. It's a format that works better at low bitrates ( < 64).

4. Just tried my .ogg files with new Audacity 1.2 pre-beta 1. They sound very well.

Thanks for your help!

[Ken Cox]

Dear Ken,

1. Please forget WMP... especially v9 !

2. When you compress a sound file, if possible start always by a .wav file, not by an already compressed format like mp3.

3. Try lower bitrates converting to OGG format. It's a format that works better at low bitrates ( < 64).

4. Just tried my .ogg files with new Audacity 1.2 pre-beta 1. They sound very well.

Thanks for your help!

Please forget WMP... especially v9 !

i luv it -- i use the sine wave patterns etc for rough comparisons when i want to try something

4. Just tried my .ogg files with new Audacity 1.2 pre-beta 1. They sound very well.

just dowloaded the 1.2 beta and it plays my oog test file without crashing

always wear my tin hat when playing --- get better resonance

k

[Guest guru]

"i luv it [WMP] -- i use the sine wave patterns etc for rough comparisons when i want to try something"

Ken, Win Media Player sounds quite bad mp3's, and its wave pattern is very vague.

Rather use the Nero Editor spectrograms, much more accurate and precise.

[alrobin]

I find, too, with the new "Audition" (formerly Cool Edit Pro), one can zoom in to the actual sine-wave components of a sound, and pin-point them accurately to the nearest thousandths of a second. Very cool!

[Guest guru]

Al, don't forget free Audacity too (see the last v1.2.0 pre-1, much improved) allows a similar zooming, theoretically up to 1/1.000.000 of a sec.!

[alrobin]

Yes, the new Audacity looks pretty good, especially for a free program.

In Audition, too, one can actually zoom in even further than thousandths of a second - like Audacity, right in to the actual samples themselves, which can be "plucked" and moved up and down to change the amplitude for a particular sample. Very cool!

However, in Audition, the main timing indicator is in thousandths, about as accurate as you really ever need to be.

[think(box)]

Guido, regarding your request of six days ago from within this topic thread, I have confirmed an OGG only timing bug:

Thanks Peter, Ralph and Bill. Reproduced the same test, with identical issue.

Now I ask you to make another test.

Michel wrote me signaling an abnormal behaviour of synchronized shows using .ogg background files. I made a little test, and I think there is another bug in 4.10-4.11.

Please make a little simple show with 10-15 random pictures, add a little .ogg music file, place automatically transition points (without transition effects is better) and synchronize exactly one slide or two (any slide) with a beat of music.

Make .exe and save .pte file.

Now reopen .pte file, and verify the synchronization in the time line, starting not only by the beginning but also by any point before the synchronized slide...

And then please tell me something about the results. Thanks!

Using the Nutcracker OGG file that you sent to me I set up the show as requested with the one precisely defined transition amidst the rest automatically positioned.

Timing was OK when making the show (in segment playback of Custom Synchronization window) but when I played the .EXE show afterward it was not the same. The slide that was supposed to appear with the cymbal crash was exactly ONE second LATE. Then I re-opened the show in PTE and found buggy behavior:

When playing from start, the cymbal crash happened at 17.3 seconds on the timeline. Yet when I previously created the show the cymbal and defined transition were at 18.3 seconds. It was as if the music was now playing ONE second advanced in time. Then when I tried playing just a segment from before my 18.3 second sync point the timing of the music CHANGED to cymbal crash at 18.3 seconds once again.

You might call this a moving target timing bug.

I would also like to point out a very short sound skip at the beginning of the show playback. The new music player is skipping where the old method did not. This has been mentioned in other threads.

My system is Win98SE, P2 400MHz, 512MB ram.

Since we should eliminate the lieklihood that we were using a defective OGG sound file, I set up a second experiment with a totally different music source file in OGG format, and I used CBR. Conversion to OGG was by the latest dBpowerAMP converter, to 96kbps. The result: Exactly the same one second bug.

Finally, I made a matching music file in MP3, 96Kbps and ran the same test. From MP3 format the playback timing was flawless under all circumstances, including before creating, after creating, segment playback or playback from the start. It didn't skip at the beginning, whether OGG or MP3 for this other music selection. The skipping problem is generic, and not so much related to sound format.

By the way, using the MP3 of exact same music I was able to determine that the earlier of the two times was the correct one. So in Guido's music test the cymbal at 17.3 seconds was correct. It was incorrect whenever it happened at 18.3 seconds.

Guido, did you find the same one second error value in the timing that you observed? Thanks

[Guest guru]

Bill, I thank you very much for your tests. The results of mine ones are identical to yours: 1 second error, in the same situations.

But there is a sentence of your message that I didn't well understand (forgive me, my English knowledge is very poor...), when you say

"It didn't skip at the beginning, whether OGG or MP3 for this other music selection. The skipping problem is generic, and not so much related to sound format".

Thanks again.

[think(box)]

You are welcome, Guido. The failure is clear and ready for Igor to debug.

Regarding your march-end.ogg sound file I think it just has a rough start. Even dBpowerAMP doesn't start very well on that audio file. I did a lot more testing before posting this reply.

This is a long one, but here is the sound performance understanding that I have today:

The skipping performance issues are not quite so clear, nor is it easy for Igor or *anyone* to code around the sound system hardware limitations. I believe the fundamental cause of skipping and the place to properly "fix" it is in sound card data buffers. They simply should be larger, so that Windows can be sloppy with the sound data stream and yet not cause skips.

But that is not the way sound hardware is today in millions of systems out there. We're stuck with undersized hardware audio data buffers in sound cards. Even eight years ago the industry was using 1MB in video cards. If they used just 1MB buffers in sound cards, there would be a seven second sound buffer duration, totally eliminating skips. There generally hasn't been enough market pressure to justify the cost of larger hardware buffers in sound cards.

So I wish to declare that I just don't have sufficiently complete, validated and repeatable failure information regarding RELATIVE performance of P2E audio when compared to Windows Media Player audio, using WMP v9 as the metric. WMP v9 doesn't repeat play the sound buffer data when it skips, while P2E does repeat. Hence WMP and P2E skip differently, but they both DO skip under system CPU load stress.

Some of the other forum topics and replies have taken shots at the skipping issue, but do we have conslusive test results? I want to look at some nasty stress-test a/v cases that I have for CD-ROM based shows, but haven't yet built them under v4.11 and loaded them to a CD for playback tests.

Under P2E v3.80 I had zero trouble with sound playback, playing entire very large shows from CD (over 600MB). P2E v4.01 had a "slow loading" problem with my very large a/v shows, so I kept them at v3.80.

P2E v4.11 has totally fixed load time issues, as has been posted. But there is a fundamental difference with v4.11 sound playback. The sound data is no longer copied to hard drive for WMP playback as it was in earlier P2E shows.

I have simply had so much fun using the new v4.11 features and working out dramatically more interesting show designs that I haven't covered the basic sound performance tests sufficiently. I know that I have more skipping than before, but I do not understand it well enough to give Igor a specific case to debug. And I have tested stressful shows ONLY from hard drive. I have placed small, low-stress shows on CD and they do work correctly.

Perhaps one good test is to take all of the Beechbrook-posted shows made since v4.11 and load the whole bunch on a CD for playback testing. I've only played them from hard drive so far.

[Guest guru]

Thank you Bill for your kind and very clear explanation.

From your tests with 3.80/4.0 and 4.10/4.11 it actually seems a question of sound buffering.

[alrobin]

That was the problem in making PTE 4.1 compatible with mp3 on Win2K, dual-processor mode.

[think(box)]

I finally locked down a skipping & timing bug with a repeatable demo and have sent its zip archive to Igor for study. This one happens when an MP3 track starts with an image. There may be only one skip/timing bug(?!)

Igor needs demo cases with repeatable failures so that he can be sure all is OK.

Al, this one was found on Win98SE, FYI.