Topic: speed correction

[OOK]

Does anyone know the difference in semitones and cents there is between a dat recorded at 44.1 vs. 48.  I downloaded a show that sounds great.....however it sounds slow as if when it was transfered to the computer from dat there was a mismatch in sample rates causing the recording to sound slow.........not wanting to delete the  download I took it as a challenge to try a correct the speed in SF6.......................
I believe the difference to be +1 semitone and +36cents............

The recording in question is DBT3-5-05 the 421>D8 source.........I am familiar with DBT and know Pattersons voice well, so in correcting the sound of his voice all else seems to fall in place............   any knowledge or suggestions..............?

[OOK]

bump?

[pfife]

can you change the sample rate header of the file in an editing program?  Otherwise, I am of no more suggestions... sorry man...

[cleantone]

Thats what I was gonna say to try. I've never done it but you might be able to change the head to read properly and the problam may be solved without the pitch shift processing. Soundhack is the freeware I use for header stuff.

[leegeddy]

try:  44.1/48 which is approximately 0.91875.  see if you can squeeze the show to that amount. 

marc

[OOK]

actually I think you did the math backwards...I think I should be 48/44.1  (assuming the original file was 48K which I think it was).....but I could be wrong...which gives you 1.088435
which would equate to +1 semitone and rounding up +9 cents......I tried that but it still sounds a little slow......But using my ears it sounds good at +1semitones and +36 cents..............I tried everything and shifting the pitch is the only thing that seems to work............................

[cleantone]

So you tried changing the header to 48khz? I'm really curious to see if this would work. I've done this in the past and in case it ever should happen again I'm very curious.

[leegeddy]

actually I think you did the math backwards...I think I should be 48/44.1  (assuming the original file was 48K which I think it was).....but I could be wrong...which gives you 1.088435
which would equate to +1 semitone and rounding up +9 cents......I tried that but it still sounds a little slow......But using my ears it sounds good at +1semitones and +36 cents..............I tried everything and shifting the pitch is the only thing that seems to work............................

assuming that the original wav was at 48k sampling and the person transferred the show at 44.1k, the 91.875% represents the difference.  

if you have CoolEdit or similar wave editor, try applying the 91.875% number and run the TIME/PITCH application. preserver neither the pitch and the tempo.

see what the results would be.

marc

[OOK]

I use SF6  and it doesn't give me pecentages. wait yes it does....but the time  and pitch are two seperate applications....I used the pitch App...... It measure in semi-tones and cents..........I am sure I have it work out now.  I went with +1semitone and +36 cents.   Patterson voice sounds as it should and the rest fell into place.....Thanks for the help............

[SparkE!]

I use SF6  and it doesn't give me pecentages. wait yes it does....but the time  and pitch are two seperate applications....I used the pitch App...... It measure in semi-tones and cents..........I am sure I have it work out now.  I went with +1semitone and +36 cents.   Patterson voice sounds as it should and the rest fell into place.....Thanks for the help............

Actually, you need to adjust by 1.467 semitones.  I'm assuming that your definition of a semitone is one half step, like the difference between the note C and the note C#.  One semitone up is the original frequency times 2^(1/12) = 1.05946.  The ratio you are off is 1.0884.  So you take log(1.0884)/log(1.05946) and you get 1.467 which is the number of half steps (or semitones) you need to correct.

[OOK]

Thank you!!!!!!now I am off 10.7 cents I will adjust!

[dklein]

Before you do all that destructive processing, what happened to the original suggestion of just changing the sample rate back to 48k - that should fix the pitch and make it play in real time.  Then if you want to burn to CD, you can resample yourself.
(changing sample rate only take a second - there should be no processing involved).

[SparkE!]

Both methods of resampling should produce comparable results, but I agree with dklein that it is desirable to correct the sampling rate of the original since that is the error that was originally made.  This does not change the samples themselves.  It just changes the sample rate that is reported by the wav file. Once that is done, resample away.  You can do it by adjusting semitones or by resampling to a different sampling rate.  Either way, you're doing exactly the same thing.

[cleantone]

You can do it by adjusting semitones or by resampling to a different sampling rate.  Either way, you're doing exactly the same thing.

Are you sure about that? I can't say I am certain but I wouldn't think a resampling process is the same as digital pitch shifting. I would think that the pitch shift would be more destructive and potentially hurt the sound much more than resampling 48 to 44.1. Anyone elses thoughts?

[SparkE!]

Hey cleantone, I think you're probably correct to question my statement.  I think that SF will just shift pitches and not change the length of the file.  In this case you want both and the resampling is probably the way to go.  Thanks for the well placed dope slap!