If you feed an encoder set for 44 an input running at 48, what are the ramifications other than pitch shift? Will the encoder start skipping as well?
Thanks.
Thanks.
Digital guru needed
- Thread starter Dick Rees
- Start date
Re: Digital guru needed
If you are asking about sending a digital platform running internally at a 44kHz sample rate, 48k samples of data per second. there will be 4,000 samples extra each second, not to mention the per sample timing will not line up nicely. Hypothetically if the digital platform could magically ignore those extra 4,000 samples, how it does that will impact whether there is a pitch shift or not.
Say the platform grabs just the first 44,000 samples of the 48,000 each second, and stretches them out as they came in to fill the full one second, you would hear a click each second caused by the 4k sample discontinuity each second, and a slight pitch shift (lower). If the 44k platform resampled the 48k data on the fly and just grabs the closest 48k input sample data to put into each 44k sample time hole, the sound will not experience a pitch shift or large perturbations. (this is sort of what sample rate conversion does).
To perform pitch shift on purpose by reading data in at one sample rate and reading that same date out at a different sample rate, you either end up with too much data that you need to discard some fraction of, or not enough data creating holes in the real time output. To discard data there are a number of sample splicing algorithms to reduce perturbations, to fill holes one common technique is to repeat redundant information again with careful splicing.
JR
Encoder? The typical ramification is that you will get no output.
If you are asking about sending a digital platform running internally at a 44kHz sample rate, 48k samples of data per second. there will be 4,000 samples extra each second, not to mention the per sample timing will not line up nicely. Hypothetically if the digital platform could magically ignore those extra 4,000 samples, how it does that will impact whether there is a pitch shift or not.
Say the platform grabs just the first 44,000 samples of the 48,000 each second, and stretches them out as they came in to fill the full one second, you would hear a click each second caused by the 4k sample discontinuity each second, and a slight pitch shift (lower). If the 44k platform resampled the 48k data on the fly and just grabs the closest 48k input sample data to put into each 44k sample time hole, the sound will not experience a pitch shift or large perturbations. (this is sort of what sample rate conversion does).
To perform pitch shift on purpose by reading data in at one sample rate and reading that same date out at a different sample rate, you either end up with too much data that you need to discard some fraction of, or not enough data creating holes in the real time output. To discard data there are a number of sample splicing algorithms to reduce perturbations, to fill holes one common technique is to repeat redundant information again with careful splicing.
JR
Re: Digital guru needed
Back in the early '70s I worked as a technician for a company developing time-compression or pitch shift for pre-recorded material, predominantly targeting talking books for blind market. Since you can comprehend pre-recorded speech at 2x or more if pitch corrected. Of course there were licensees working on small amounts of time compression with pitch correction for squeezing an extra commercial or three into a late night movie.
As the opposite of ignorance being bliss, I am now cursed with knowing how to recognize the sound of old movies being time squeezed. Hint- it is most audible on orchestral background music. where the perturbations caused by splicing between samples to discard the now too much data, causes a burbling sound (technical term) with a period in the half second to 2 second repeat rate. Since they seem to already play 2 hour movies in 3 hours of TV time this wouldn't seem necessary. Another place this is apparent is in old serial TV shows produced for tight 30 min or 60 min slots, while some satellite channels don't even pretend and put 30 min shows in 40 min slots, or even play several in a row (like MASH) with different length slots for each.
This discussion reminded about all this, and now I'm freshly sensitive to this again. I'm a little surprised that they aren''t using newer different technology, not there are any completely transparent ways to seamlessly discard X% of the data without hearing something. Actually this is not very audible on speech due to percussive nature and short pitch periods (singing with longer held notes is harder). It's mostly on the orchestral passages holding chords for several seconds where you can really hear it burbling away.
Thanks Dick...
JR
Back in the early '70s I worked as a technician for a company developing time-compression or pitch shift for pre-recorded material, predominantly targeting talking books for blind market. Since you can comprehend pre-recorded speech at 2x or more if pitch corrected. Of course there were licensees working on small amounts of time compression with pitch correction for squeezing an extra commercial or three into a late night movie.
As the opposite of ignorance being bliss, I am now cursed with knowing how to recognize the sound of old movies being time squeezed. Hint- it is most audible on orchestral background music. where the perturbations caused by splicing between samples to discard the now too much data, causes a burbling sound (technical term) with a period in the half second to 2 second repeat rate. Since they seem to already play 2 hour movies in 3 hours of TV time this wouldn't seem necessary. Another place this is apparent is in old serial TV shows produced for tight 30 min or 60 min slots, while some satellite channels don't even pretend and put 30 min shows in 40 min slots, or even play several in a row (like MASH) with different length slots for each.
This discussion reminded about all this, and now I'm freshly sensitive to this again. I'm a little surprised that they aren''t using newer different technology, not there are any completely transparent ways to seamlessly discard X% of the data without hearing something. Actually this is not very audible on speech due to percussive nature and short pitch periods (singing with longer held notes is harder). It's mostly on the orchestral passages holding chords for several seconds where you can really hear it burbling away.
Thanks Dick...
JR