| Stereo Tool https://forums.stereotool.com/ |
|
| Something like "Spectral Band Replication" https://forums.stereotool.com/viewtopic.php?t=7271 |
Page 1 of 1 |
| Author: | near05 [ Mon Jan 11, 2016 1:19 pm ] |
| Post subject: | Something like "Spectral Band Replication" |
Hello there, I have the following suggestion: Many old recordings have missing high frequencies (> 12000 Hz). This concerns, for example, music cassettes. Over time, the high tones are lost. Now there are methods such as "Spectral Band Replication". This calculates the pitch based on the existing audio. As a result, you hardly hear it. I understand that this data, however, are collected from existing high frequencies, so you can calculate this based on the fundamental frequencies. It would be nice if there would be such a functionality for existing audio, without the high frequencies. Most high frequencies would have to be able to calculate well on the basis of the fundamental frequencies. So you would have old recordings can breathe a little more life. |
|
| Author: | hvz [ Mon Jan 11, 2016 1:51 pm ] |
| Post subject: | Re: Something like "Spectral Band Replication" |
I've been thinking about this (in fact, True Bass is something like this, but for bass). Will look into it. |
|
| Author: | hvz [ Mon Jan 11, 2016 2:24 pm ] |
| Post subject: | Re: Something like "Spectral Band Replication" |
Hm, that kinda works but it sounds a bit 'shrill' on S sounds. For hi-hats it works. Will keep trying  Edit: Good I think I just figured out how to determine whether a sound will be shrill or not. It's kinda similar to rumbling in True Bass. You don't want almost no boost of highs at the frequency where you start regenerating the signal, and a lot more at higher freqs.
|
|
| Author: | near05 [ Mon Jan 11, 2016 8:10 pm ] |
| Post subject: | Re: Something like "Spectral Band Replication" |
That sounds excellent. Then it works even with Brass Sounds? These sounds always have certain overtones which are very well audible. I'm really looking forward to test it, if you have done a version of it. If this really works, it would be incredibly awesome! 
|
|
| Author: | hvz [ Tue Jan 12, 2016 12:44 am ] |
| Post subject: | Re: Something like "Spectral Band Replication" |
Well I think it sounds bad for brass sounds, and that's one of the things that I want to filter out.. Maybe I shouldn't, not really sure yet. It works for hi-hats though. |
|
| Author: | near05 [ Tue Jan 12, 2016 5:56 pm ] |
| Post subject: | Re: Something like "Spectral Band Replication" |
That does not sound so good. In my eyes, this information should be available on each case. Perhaps you would have to make a detailed harmonic analysis to calculate the overtones. I've sent you via PM a sample file. This contains a set of drums, and a few other instruments. Could you please listen at this opportunity once, and apply your algorithm on it? There are always parts of strings / brass, which is actually expected even after 11025 Hz. Thanks in advance! |
|
| Author: | hvz [ Tue Jan 12, 2016 9:06 pm ] |
| Post subject: | Re: Something like "Spectral Band Replication" |
I have just sent you the processed file via a PM. Let me know what you think (you can do that here - I won't post samples here publically before I'm really done). |
|
| Author: | Modulator [ Wed Jan 13, 2016 1:05 pm ] |
| Post subject: | Re: Something like "Spectral Band Replication" |
Oh yes! This seems (not sounds, yet) wonderful!
|
|
| Author: | near05 [ Wed Jan 13, 2016 6:12 pm ] |
| Post subject: | Re: Something like "Spectral Band Replication" |
Quote: I have just sent you the processed file via a PM. Let me know what you think (you can do that here - I won't post samples here publically before I'm really done).
That sounds great! I think the points with the brass very well. Now the whole piece sound much more alive. However, I still noticed something with the tambourine. Unfortunately, here the completion has not worked quite. The high frequencies are missing.The brass at the beginning: At the beginning of the completion has worked very well. But then at about 1 second, unfortunately, lack the information in the pitches. In about 8 seconds, then in the middle of the spectrum a bit to hear too much of the brass. That's not really bad, and I think most people will never notice. The In 17 seconds where the brass exactly play the same part, you won't notice it. When subtracting the left channel from the right, then you see that there is in the middle a "hole". Could you perhaps generally smooth the transition a little better? Is there the effect has the chance to complete over Left / Side perform? I think you should combine this somehow. All in all a very good job! I have only thought the worst after you had written that the brass do not sound as good. But I think that's not too bad. I look forward, if you continue to build on it. Maybe you can do a few things to optimize, and then I re-render of the piece a version. Great Work!  
|
|
| Page 1 of 1 | All times are UTC+02:00 |
| Powered by phpBB® Forum Software © phpBB Limited https://www.phpbb.com/ |
|