Start a new topic
Solved

waveform != waveform

I don't know what Neon is showing as "waveform-slider", but there's definitive no waveform...


Comparing the waveforms Audacity shows with whatever Neon shows, I can only get to one conclusion: Neon does not show a waveform.


Audacity (left channel), Neon, Audacity (right channel)


Neon placed above each channel from Audacity

Should be since it uses this data:

http://www.bass.radio42.com/help/html/6a20df3a-a6fd-486b-74e7-3a0a65f7e545.htm


Note: It's up to each presentation device to how they present data, Auda city for example might use enhancement algorithms and different scales so I assume there's no general right or wrong.

If you look at the main "form" it is the same.

FFT kann unterschiedlich sein abhängig von der Berechnung
FFT / FFTW usw...
Wir verwenden BASS_ChannelGetData and BASS_WASAPI_GetData
FFT von BASS_ChannelGetData(stream) ist nicht das gleiche wie FFT vom BASS_WASAPI_GetData(Device)

Keine Ahnung was Audacity verwendet.

Es kann nicht gleich sein.

----------------

FFT can be different depending on the calculation
FFT / FFTW etc ...
We use BASS_ChannelGetData and BASS_WASAPI_GetData
FFT of BASS_ChannelGetData (stream) is not the same as FFT from BASS_WASAPI_GetData(Device)

No idea what Audacity uses.

It can not be the same.


also no BUG


greets


Just as Emil says, the FFT can be calculated in various ways. Possibly AudioCity is using the Nyquist transform algorithm as an extra steps (not sure what the light-blue peaks in AudioCity represent, but it might indicate a discrete transformation and/or some modulation).


But depending on which FFT transformation thats choosen, scaling and possibly modulation a waveform can look different.

Worth to mention is that that AudioCity seems to render with some anti-aliasing which also gives a different look. In Neon the interpolation is applied to the data selected instead.


So, in short, no bug.

And..

Audacity is not the measure of all things ;) (etc.. Great Soft for analyze Audio)


if you want compare then use a Application which used the Bass Library (Bass.NET)

Thanks for the Report


greets

If I add a baseline, you can see that it's not the same, especially at the beginning.


Can bass create a waveform for 0.1 seconds in 2000x400 pixels and save it into a png-file?
For example create a waveform from position 1m00,000 to 1m00,100.

>>The WaveForm will be drawn as two independent stereo envelopes, a combined dual mono envelope
>>or a mono envelope.
Can you add an option to get "two independent stereo envelopes"?
Natürlich weiß ich, dass es auch darauf ankommt, wie man die "durchschnittlichen" Werte für einen gewissen Zeitraum berechnet, aber die Ergebnisse sollten sich schon ein wenig ähneln, und diese weichen doch recht deutlich von einander ab... Daher meine Frage nach einem kleinen Zeitraum auf einem großen Bild. ;-)

Wenn ich mir das Bild mit den ergänzten Baselines ansehe, dann schein mir da eine Art Mixtur aus beiden Kanälen gezeichnet zu werden, oben die positiven Werte des linken, unten die negativen Werte des rechten Kanals. Eine sehr merkwürdige Art zwei Kanäle anzuzeigen.

Falls ich damit richtig liege, dann kann ich nur sagen, dass man deutlich bessere Ergebnisse bekommt, wenn man zwei separate Wellenformen anzeigt - sogar wenn man nur eine einzelne Monospur anzeigt, sind die Ergebnisse besser.

-----

I know there may be some differences in calculating the "average" value for a time period, so the results may vary, but both "waveforms" are not even close, that's why I asked for a very small part of the track with a huge pixel-count.

If I take a look at the picture with the added baseline, I think it might be some mixture of both channels, the upper half showing the positive values of the left channel, the lower half showing the negative values of the right channel - a really strange way to show two channels.


If I'm correct with this, getting the waveform drawn as two independent stereo envelopes will create much better results. Even getting a single mono waveform would be better.

>> (not sure what the light-blue peaks in AudioCity represent, but it might indicate a discrete
>> transformation and/or some modulation)

Quoting "http://manual.audacityteam.org/o/man/audacity_waveform.html":
>> The dark blue part of the waveform displays the tallest peak in the area that pixel represents. At
>> default zoom level Audacity will display many samples within that pixel width, so this pixel
>> represents the value of the loudest sample in the group.
>> The light blue part of the waveform displays the average RMS (Root Mean Square) value for the same
>> group of samples. This is a rough guide to how loud this area might sound, but there is no way to
>> extract or use this RMS part of the waveform separately.

>>If I'm correct with this, getting the waveform drawn as two independent stereo envelopes will create much better results. Even getting a single mono waveform would be better.

Please add this as a feature request, it should be possible to solve I think.

The current look is a combined look as earlier mentioned which shows a combination of the left and right FFT peaks.

(I assume it will be easier to implement than the spectogram where I just got stuck in the FFT transformations :-o)

>>The current look is a combined look as earlier mentioned which shows a combination of the left and right >>FFT peaks.


I think also..


>>Falls ich damit richtig liege, dann kann ich nur sagen, dass man deutlich bessere Ergebnisse >>bekommt, wenn man zwei separate Wellenformen anzeigt - sogar wenn man nur eine einzelne >>Monospur anzeigt, sind die Ergebnisse besser.


@Sven M.

Das ist schon richtig.. :)

Könnte man in der Zukunft noch ändern.

Ist halt immer das selbe leidige Problem wenn man es nicht selber macht sondern fertige Classen / Controls oder was auch immer verwendet.


greets

>>>>Falls ich damit richtig liege, dann kann ich nur sagen, dass man deutlich bessere Ergebnisse >>>>bekommt, wenn man zwei separate Wellenformen anzeigt - sogar wenn man nur eine einzelne >>>>Monospur anzeigt, sind die Ergebnisse besser.

>>

>>@Sven M.

>>Das ist schon richtig.. :)

>>Könnte man in der Zukunft noch ändern.

>>Ist halt immer das selbe leidige Problem wenn man es nicht selber macht sondern fertige Classen

>>Controls oder was auch immer verwendet.


Naja, die angegebene Klasse kann ja drei Varianten:


>>The WaveForm will be drawn as two independent stereo envelopes, a combined dual mono envelope

>>or a mono envelope.


----


The mentioned class should be able to create my requested variants, so this should be quite easy to implement.


1 person likes this

>>The mentioned class should be able to create my requested variants, so this should be quite easy to implement.


Please note that BASS.NET is used to generate the peak data, I'm doing the rendering myself since the Bass.net class did not support what I needed (hover WPF, gradient) so that's something I will need to do manually, hence the need for a separate feature request.

OK, I see. I will create a feature request.

 

Login or Signup to post a comment