Re: sample rates in audio
Theoretical only... on recordSound set the recordsamplesize to 16 set the recordrate to 2.5 set the recordformat to wave record sound file gAudioTestPath end recordSound Doesn't do anything... movie props still show 16 bit, 8 kHz... and files is exactly the same as on recordSound set the recordsamplesize to 11 set the recordrate to 5 set the recordformat to wave record sound file gAudioTestPath end recordSound on recordSound set the recordsamplesize to 16 set the recordrate to 8 set the recordformat to wave record sound file gAudioTestPath end recordSound each of the above delivers the same size file, quality and movie props in QT pro I don't get anything decent with 8 bit at all. On May 17, 2006, at 12:41 PM, Stephen Barncard wrote: Hey if 2.5 is ok for your high freq response - fine for speech and some sounds. Music? hm... sqb Oh my... now wonder it is so confusing... recordSampleSize is NOT the sample rate... On May 16, 2006, at 3:15 PM, Stephen Barncard wrote: I don't know how rev would store an 11 bit wide word without waste. Computers store info in their nice 8-wide world. It comes down to parts and sanity. There are packing methods to do this - is that's what's happening here? Try a test. See if there is no difference in file size between using 11 bit and 16 bit wide for RecordSampleSize Confirmed, you are so right... on recordSound set the recordsamplesize to 11 set the recordrate to 5 set the recordformat to wave record sound file gAudioTestPath end recordSound 12 secs -- 184K open in QT Pro, get movie props... it shows 16 bit, 8 khz on recordSound set the recordsamplesize to 16 set the recordrate to 8 set the recordformat to wave record sound file gAudioTestPath end recordSound 12 secs -- 184K Same size, same quality... Sivakatirswami -- stephen barncard s a n f r a n c i s c o - - - - - - - - - - - - ___ use-revolution mailing list use-revolution@lists.runrev.com Please visit this url to subscribe, unsubscribe and manage your subscription preferences: http://lists.runrev.com/mailman/listinfo/use-revolution ___ use-revolution mailing list use-revolution@lists.runrev.com Please visit this url to subscribe, unsubscribe and manage your subscription preferences: http://lists.runrev.com/mailman/listinfo/use-revolution
Re: sample rates in audio
Oh my... now wonder it is so confusing... recordSampleSize is NOT the sample rate... On May 16, 2006, at 3:15 PM, Stephen Barncard wrote: I don't know how rev would store an 11 bit wide word without waste. Computers store info in their nice 8-wide world. It comes down to parts and sanity. There are packing methods to do this - is that's what's happening here? Try a test. See if there is no difference in file size between using 11 bit and 16 bit wide for RecordSampleSize Confirmed, you are so right... on recordSound set the recordsamplesize to 11 set the recordrate to 5 set the recordformat to wave record sound file gAudioTestPath end recordSound 12 secs -- 184K open in QT Pro, get movie props... it shows 16 bit, 8 khz on recordSound set the recordsamplesize to 16 set the recordrate to 8 set the recordformat to wave record sound file gAudioTestPath end recordSound 12 secs -- 184K Same size, same quality... Sivakatirswami - which is NOT the sample rate by the way - it is the width of the word describing the voltage of the waveform at that slice of time) -- the sample rate is actually set by the recordRate property typical Rev misleading property and command names. At least there's no DestroyRecording command. If I used the sound features I would immediately bounce the params from my own well-named commands and call them what I like. Also define with standard rates in constants. sqb ___ use-revolution mailing list use-revolution@lists.runrev.com Please visit this url to subscribe, unsubscribe and manage your subscription preferences: http://lists.runrev.com/mailman/listinfo/use-revolution
Re: sample rates in audio
In my recent researches into wav and aiff formats I learnt that these formats allow for pretty much any word length - but as you say, they get stored as the next multiple of eight, padded with zeros. Maybe there's some issue of A to D and D to A conversion or latency that makes it worthwhile to use 11 bits? Given that each additional bit doubles the resolution, maybe there are worthwhile trade-offs involved. Best, Mark On 17 May 2006, at 02:15, Stephen Barncard wrote: I don't know how rev would store an 11 bit wide word without waste. Computers store info in their nice 8-wide world. It comes down to parts and sanity. There are packing methods to do this - is that's what's happening here? ___ use-revolution mailing list use-revolution@lists.runrev.com Please visit this url to subscribe, unsubscribe and manage your subscription preferences: http://lists.runrev.com/mailman/listinfo/use-revolution
Re: sample rates in audio
Hey if 2.5 is ok for your high freq response - fine for speech and some sounds. Music? hm... sqb Oh my... now wonder it is so confusing... recordSampleSize is NOT the sample rate... On May 16, 2006, at 3:15 PM, Stephen Barncard wrote: I don't know how rev would store an 11 bit wide word without waste. Computers store info in their nice 8-wide world. It comes down to parts and sanity. There are packing methods to do this - is that's what's happening here? Try a test. See if there is no difference in file size between using 11 bit and 16 bit wide for RecordSampleSize Confirmed, you are so right... on recordSound set the recordsamplesize to 11 set the recordrate to 5 set the recordformat to wave record sound file gAudioTestPath end recordSound 12 secs -- 184K open in QT Pro, get movie props... it shows 16 bit, 8 khz on recordSound set the recordsamplesize to 16 set the recordrate to 8 set the recordformat to wave record sound file gAudioTestPath end recordSound 12 secs -- 184K Same size, same quality... Sivakatirswami -- stephen barncard s a n f r a n c i s c o - - - - - - - - - - - - ___ use-revolution mailing list use-revolution@lists.runrev.com Please visit this url to subscribe, unsubscribe and manage your subscription preferences: http://lists.runrev.com/mailman/listinfo/use-revolution
Re: sample rates in audio
Amazing... Stephen... I really don't know much about this... I mean I don't know theactually audio physics between bits and sampling rates. Of course sample rates seems obvious... how often the recorder is trapping for sound, but how that relates to 'bits is mysterious... anyway... just going with what you said: on recordSound set the recordsamplesize to 11 set the recordrate to 5 set the recordformat to wave record sound file gAudioTestPath end recordSound I'm getting really decent quality at about 15 K per second or 900 K per minute i.e. 11 seconds gives us a 168 K file... I don't think it gets any better than that! And you seem to have declared the bottom threshold... as anything less than a sample rate of 11 starts to break down noticeably... I guess everyone has figured this out long ago... Sivakatirswami On May 15, 2006, at 6:10 AM, Stephen Barncard wrote: Actually you could halve that sample rate to 11 K and still have good intelligibility. Using the Nyquest theorem, the highest frequency you can record at a given sample rate is approximately less than half its sample rate. AM broadcast stations cut off at 5khz, which can be handled easily by the 11k sample rate. This is not 11 k bits per second, like in MP3s, but a full parallel 16 bits at 11k and without compression, will have no artifacts, just bandwidth reduction - i.e. less high end. sqb ___ use-revolution mailing list use-revolution@lists.runrev.com Please visit this url to subscribe, unsubscribe and manage your subscription preferences: http://lists.runrev.com/mailman/listinfo/use-revolution
Re: sample rates in audio
I just checked out the docs. I'd never looked at this before. recordRate = 5 gives a sample rate of 5k -- THIS is the SAMPLE RATE According to Nyquest, the highest frequency that can be passed at a sample rate of 5k/second is 2.5khz. recordSampleSize = number of bits used to record = 11 -- THIS is the BIT WIDTH I am not aware one could specify odd bit lengths. The number of bits determine the dynamic range - how many steps are used to represent the high and low points in the audio waveform. This is an odd pair - you should test on several players. I was not aware that sound files rate and bit properties would be recognized on a sliding scale. I would check for the 'chipmunk effect' on voices. There are standards that are base on the power of 2 96 48 32 88.2 44.1 22.05 11.025 5.5125 should be standard values for recordRate if the hardware can do it. The fractional sample rates are that way because early CD players were based on color TV crystals for frequency stability. Because they were around already in the millions and cheap. Also video tape machines were the only devices around that had the bandwidth to record audio digitally at the time. The digital audio had to be turned into an analog video signal to record on tape! You could look at it on a video monitor. And again these numbers below are 'standard' bit widths. 32 24 16 8 -- recordSampleSize I don't know how rev would store an 11 bit wide word without waste. Computers store info in their nice 8-wide world. It comes down to parts and sanity. There are packing methods to do this - is that's what's happening here? Try a test. See if there is no difference in file size between using 11 bit and 16 bit wide for RecordSampleSize - which is NOT the sample rate by the way - it is the width of the word describing the voltage of the waveform at that slice of time) -- the sample rate is actually set by the recordRate property typical Rev misleading property and command names. At least there's no DestroyRecording command. If I used the sound features I would immediately bounce the params from my own well-named commands and call them what I like. Also define with standard rates in constants. sqb Amazing... Stephen... I really don't know much about this... I mean I don't know theactually audio physics between bits and sampling rates. Of course sample rates seems obvious... how often the recorder is trapping for sound, but how that relates to 'bits is mysterious... anyway... just going with what you said: on recordSound set the recordsamplesize to 11 set the recordrate to 5 set the recordformat to wave record sound file gAudioTestPath end recordSound I'm getting really decent quality at about 15 K per second or 900 K per minute i.e. 11 seconds gives us a 168 K file... I don't think it gets any better than that! And you seem to have declared the bottom threshold... as anything less than a sample rate of 11 starts to break down noticeably... I guess everyone has figured this out long ago... Sivakatirswami -- stephen barncard s a n f r a n c i s c o - - - - - - - - - - - - ___ use-revolution mailing list use-revolution@lists.runrev.com Please visit this url to subscribe, unsubscribe and manage your subscription preferences: http://lists.runrev.com/mailman/listinfo/use-revolution
sample rates in audio
Actually you could halve that sample rate to 11 K and still have good intelligibility. Using the Nyquest theorem, the highest frequency you can record at a given sample rate is approximately less than half its sample rate. AM broadcast stations cut off at 5khz, which can be handled easily by the 11k sample rate. This is not 11 k bits per second, like in MP3s, but a full parallel 16 bits at 11k and without compression, will have no artifacts, just bandwidth reduction - i.e. less high end. sqb OK, yes, it works... but one must stick with the default codec none ... any attempt to set it to ulaw with the USB mic doesn't work... and touching the input setting can also be a problem (i.e. dflt is what we want nothing else) But anyway, it doesn't seem to matter too muich. For voice I'm getting small file sizes at 16 bit, 22.01kHz. (8 bit introduces to much hiss and pops.) Setting the record format to wave and the file extension to .wav means it boots on both platforms which is great... WMP on Windows and QT on the Mac. [ if you want quality -- giant files... set the recordsamplesize to 128.21 set the recordrate to 44.1 ] -- stephen barncard s a n f r a n c i s c o - - - - - - - - - - - - ___ use-revolution mailing list use-revolution@lists.runrev.com Please visit this url to subscribe, unsubscribe and manage your subscription preferences: http://lists.runrev.com/mailman/listinfo/use-revolution
