Streaming V Red book CD quality - Education needed
Posted by: Southweststokie on 28 January 2012
Would someone please educate me on this business of sample rates and speeds, 192khz etc. Or point me to a good web link that will explain the difference in a language that does not require a degree in electronics to understand it. I'm just a simple mechanical engineer.
Thanks in advance for your guidance,
Ken
Someone else may explain it better, but I'll give it a quick try.
A standard music CD ("redbook CD") is 16 bit 44.1 kHz encoding/quality.
Streaming is more about the transport (storage->player->dac->amp etc.)
Streaming can be playing files that are either:
- equal to CD/redbook (16/44.1)
- superior to CD (24 bit, 96 - 192 kHz)
- worse than CD (lossy MP3s etc.)
EDIT:
So, no, redbook/standard CD is not the best quality available for music reproduction.
It's a "compromise format" dating from the early 1980s.
In terms of the sample rate and bit-depth:
When a musical sound (an inherently analogue thing) is digitised then it is done so my means of measuring it's amplitude (how loud it is) at regular intervals, and assigning that amplitude a numeric value at each interval.
The accuracy of that value, and the number of possible different values that can be assigned is dependent on the bit-depth. With a 16 bit binary code (i.e. 16 bits of information, each one either a 1 or a 0) the number of possible different values that can be assigned to the amplitude of the musical signal at any given point in time is 65,536. That partly defines the resolution of the recording - it is only possible to record 65,536 different gradations in the amplitude of the signal. This also defines to total dynamic range of the signal (i.e. the difference between the loudest and quietest sounds possible on a recording) - 96 decibels for a16 bit recording.
If you record with a bit depth of 24 bit, then that means you can record with 16,777,216 different gradations of amplitude, and it yields a dynamic range of 144db - so clearly superior; a more accurate sample of the original analogue signal at any given point in time.
The sample rate is how often this measure of the amplitude of the signal is taken. With the 44.1khz of CD, this means that the musical signal has been sampled 44,100 times per second at regular intervals. This also contributes to the overall resolution of a recording - any changes occurring between the sample points are simply missed on the recording. The sample rate also determines the highest frequency that can be recorded - with 44.1kz the highest frequency that can be recorded is 22.05khz. With higher sampling rates comes greater resolution and higher frequencies able to be recorded.
Hope that clarifies.
Your comments appreciated as always.
Thanks,
Ken
Also our hearing rarely extends to 20kHz and so Redbook has the potential of sounding extremely accurate.
The complication is that music is mastered and edited so it sounds good across a wide range of playout devices; car radio, mp3 player, tv, budget hifi, Bose speakers etci. This means the sound is often manipulated by squashing it and boosting certain frequencies so it sounds good on mediocre replay devices. CDs often carry this modified sound. Some of the current hidef downloads are of content before it is processed, so it sounds better. But if there was a Redbook quality download (16/44.1) of this Pre processed audio, it would also sound superior to a compressed version that might be available generally on CD if played on a good system. This processing is most obvious on very commercialised music targeted for radio and mp3/AAC download audiences.
The CDS3 has an extremely good DAC and filter but clearly optimised for Redbook replay only. It took a 555PS to go with my NDAC (and NDX) before I got a significantly more satisfying replay from Redbook over myCDS3 (in this case using upnp streaming as the CD transport)
Simon
all well designed amplifiera will sound different mainly by the bandwidth /gain product. each manufacture will tweak it for want they percieve is correct. as is cdp too. even worse is the units of a box speaker.take you pick. simon is correct, if the filtering are to your taste then you are happy/. high res playback does appear alittle bright because of the extended band width.
most hifi guys i better are middle aged and they rarely can hear much over 16 k hz
Thanks DT79! Brilliant succinct explanation. Shift key broken Rich?
Another couple of factors in replay from CD vs streaming to a DAC stem from the way a CD player extracts music.
In a CD player the stream of data representing the music is read in 'real time', the speed at which the music is actually playing ( takes say 45 -60 mins to read a complete CD) and while there is relatively crude buffering in CD players the opportunity to resample any data that didn't get read right the first time is very limited (why sometimes an CD will stutter at some point ). Also the noise associated with a motorised CD mechanism is generated inside the box where your relative sensitive D/A process is taking place.
For storing on a Hard disk you can read the data off the CD much faster with no limitation in principle to how often you rescan any doubtful bit information which allows effectively an error free Rip of the music ( can be checked against a database such as 'Accuraterip' for example). You can then supply this music data to a dedicated DAC in the best noise free environment and it is much more realistic to stream the much higher data density represented by 96 & 192/24 sampling which CD players aren't designed to read ( DVD & Blu ray players use different laser wavelengths to read higher density data if you can persuade someone to encode pure music disks for them) .
As far as the audibility of frequencies to the human ear is concerned, what Simon says is true in principle but that doesn't explain why music encoded at 96/24 and 192/24 has a higher perceived audio quality. There is a some opinion that what we pick up are harmonics that ARE in our ears audible range generated by the non-audible frequencies in the playback from a DAC.
regards
Geoff
Now to perceived quality of hidef, and as I type I am listening to Meet Me in London at 192/24, I am sure much of the benefit comes from the mixing and mastering at higher bandwidths with reduced rounding, jitter and quantisation errors, as well the greater Fs is to the audio bandwidth the more likely it is possible to minimise transformation errors in the DAC and filter. (both digital and analogue). Digital encoded discrete samples of a continuous (analogue) signal can only ever be an approximation to the original signal once converted back to an analogue signal, although very close with the right conditions and methods.
Simon
Now to perceived quality of hidef, and as I type I am listening to Meet Me in London at 192/24, I am sure much of the benefit comes from the mixing and mastering at higher bandwidths with reduced rounding, jitter and quantisation errors, as well the greater Fs is to the audio bandwidth the more likely it is possible to minimise transformation errors in the DAC and filter. (both digital and analogue). Digital encoded discrete samples of a continuous (analogue) signal can only ever be an approximation to the original signal once converted back to an analogue signal, although very close with the right conditions and methods.
Simon
Simon
On the CDP point would you view the correction algorithm as being as effective as re-reading the data bits multiple times as takes place during ripping?
Another interesting discussion point in the higher resolution audio is IMO the impact of bit depth vs sample frequency. I play around with ripping from LP (using a D/A converter) at home and have found that going to 24 Bit seems to have the first most significant effect. An LP rip done at 44.1/24 certainly sounds better than one done at 44.1/16. Going to 96/24 gives a further boost but doing a 96/16 rip does not come close to the 96/24 rip. Incidentally because this is a relatively crude ripping scenario compared to sophisticated studio remastering 192/24 doesn't do much other than double the files size over 96/24 to my ears.
regards
Geoff
With regard to 16 bit and 24 bit, well assuming you standardise the amplitude level of both, the accuracy of the 16bit ADC in terms of its least significant bits needs to more accurate than the 24 bit and perhaps therefore easier to get a good result at 24bit albeit at the expense of more data produced.
Simon
Simon
Hi Simon
Good point. In truth I do the A/D conversions at 192/32 into Apple Core audio in a program called Ableton Live which then allows you to export from Core Audio in whatever frequency/bit depth you wish. All the WAV files I am comparing therefore stem from the same Core audio files...which is a bit of an unfair comparison perhaps since it is all ripped at 192/32 initially.
regards
Geoff