The images and sound section.

[Steve Anderson]

Gents,

We now have an 'Images and Sound' section of the board, I have uploaded one file of 12 seconds duration which used to be on the NBTVA website.

Is 12 seconds long enough? I wouldn't have thought so. It must take a while to achieve sync. Although some software allows 'looping' of an audio track, what would be a sensable minimum duration?

Steve A.

[DrZarkov]

No, I don't think its enough, one minute would be better. The 12 sec. are only 1 MB, that is nothing in the age of ADSL, 6 Mb would be o.k.. Even with a modem it is acceptable if you want to get just one file for testing. (I remember 10 years ago downloading the 17 Mb of MacOS 7.5.5. with a 14400 Mb Modem. That took really a long time...)

I forgot: You didn't pack the file. It would reduce the seize about less than half as big!

[Steve Anderson]

Gents,

Yesterday I started to create a number of .wav files the check the frequency response of the audio playback sections of the PCs I use here. I generated 15 files each of 30 seconds, each containing a single-frequency sine wave from 1Hz to 16.5kHz generated at -3db.

When I played them back on the laptop (Compaq Presario 2247AP, very entry-level by todays standards) I was pleased that the -3db was about 5Hz. Quite impressive I thought.

At the high-frequency end I was amazed that it was -3db at only 2kHz! I checked the the equaliser was set flat and there was no other 'modifications' to the signal being performed. I thought that there must have been something wrong in the files I created. But no, examining them at Byte/Hex level showed they were spot-on.

I also noted that even at low frequencies (200Hz) the peaks were clipped. I tried three software packages to play them back, all were the same.

I repeated the tests on my desktop PC (which is a self-assembled clone using an ASRock motherboard with on-board audio) and all was fine. The output was flat from 1Hz to 16kHz! (I didn't go any higher, but will) And no clipping of the waveform.

Lesson...check your soundcard or motherboard before you can trust it. The same will apply to the record section.

I will upload these files if there is any interest, but first I need to clean them up a little.

Steve A.

Afterthought...There's probably an equalizing network prior to the output jack to offset the silly little 20mm speakers put in laptops, hence the falloff over 2kHz.

[Steve Anderson]

For those on dial-up connections it's probably useful to compress losslessly the .wav files. The most well known is probably WinZip, but there's also WinRar and others.

Any particular preference amongst members?

This could also apply to any large file uploaded to the board.

Steve A.

[Dave Moll]

Any particular preference amongst members?

ZIP - to which I am already geared.

[M3DVQ]

lots of people already have winzip, so zip is their preference, but .RAR has a much better compression ratio IIRC.

[Klaas Robers]

My experience is that NBTV.wav files are not so much compressed by zip algorithms. Because the sound is loud and unpredictable as audio the routines have little grip onto it. Only the right channel if no sound is recorded can be compressed, giving a compression of about 50% overall.

[gary]

Yes, compressing NBTV wave files is an interesting problem, because you have video on one channel and audio on the other, both really requiring their own compression scheme. Probably the best freely/readily available compression method would be the lossless compression codecs like flac, tak, monkey's audio, etc. These should consistently provide compression down to 60% of the original with no artifacts.

However, it would be possible to write an application that compressed the audio channel separately to the video channel and thus produce much higher levels of compression by using lossless compression on the video and lossy (say mp3) on the audio.

If there was enough interest in something like this I could produce something fairly quickly.

[Steve Anderson]

It seems the the .zip format is the most widely accepted, of the NBTV .wav files I have 'zipped', most of them come out at around 50% of their original size.

This is probably better than developing a unique format for NBTV. I only suggested compressing files to help those on dail-up connections waiting forever to download them.

As Andrew has mentioned, at this stage we have plenty of storage space on this board.

Steve A.

[Klaas Robers]

Yes indeed, 50% of their original size. I suppose that is with the right channel (sound) filled with absolute silence. Those zeros can be compressed easily to almost nothing, the video channel then remains almost uncompressed.

[Steve Anderson]

The files I have created so far are mono, i.e. just video. What I am working on at the moment is a reduction of 50% by only using 8-bit data instead of 16-bit which is enough for NBTV. Even professional video data is only 10-bit.

At the usual CD sampling rate of 44.1kHz a 60 second mono file results in a file of 2.7Mbytes using 8-bit resolution. Not many software based recorders support 8 bit files, but all the playback systems I have tried so far do. Play back the Microsoft 'ding' or 'chord' .wav files from Win 3.1.

I hope to have a batch of files ready in a few days, consisting at this stage of either test waveforms, test patterns or stills.

Steve A.