Become a Site Supporter and Never see Ads again!

Author Topic: file size hypothetical theoretical question  (Read 5366 times)

0 Members and 1 Guest are viewing this topic.

Offline easyed

  • Trade Count: (6)
  • Taperssection Member
  • ***
  • Posts: 277
  • get your Jam in the Can
file size hypothetical theoretical question
« on: April 14, 2015, 09:07:29 AM »
your thoughts please? hypothetical theoretical question:

same recorder, same bit depth, same sample rate, 4 recordings, all *EXACTLY* one minute long samples

-live loud rock and roll
-ambient sound in a normal room with no music or talking
-no input (not quite silence but instead the noise floor of the recorder)
-a sine wave

question is would the file size be the same on all four recordings?

my thinking is they'd all be the same file size, in fact, i'd posit that if you used a different recorder but the same bit depth, same sample rate and the same recording examples above the file sizes would be extremely close.  if i'm wrong help me understand why, please and thanks in advance!
Beyerdynamic CK-930s > Naiant Tinybox or Littlebox > Sony PCM-M10 or
DPA 4061's > Core Sound Battery Box > Sony PCM-M10 or
matrix: Sound Devices 744T or
multitracking: Audient ASP008 preamps > JoeCo Blackbox BBR1B

Offline opsopcopolis

  • Trade Count: (3)
  • Needs to get out more...
  • *****
  • Posts: 2144
Re: file size hypothetical theoretical question
« Reply #1 on: April 14, 2015, 11:37:56 AM »
I believe it should be.  File size is not contingent on content, rather bit depth and sample rate.  Even if you use a higher bit depth/sample rate and record a lower quality sound your file still has that extra headroom whether it is being used or not

Offline Gutbucket

  • record > listen > revise technique
  • Trade Count: (15)
  • Needs to get out more...
  • *****
  • Posts: 15698
  • Gender: Male
  • "Better to love music than respect it" ~Stravinsky
Re: file size hypothetical theoretical question
« Reply #2 on: April 14, 2015, 12:01:11 PM »
Non-data compressed (WAV) files will all be identical size, regardless of their actual content.  Doesn't mater if the content is recorded silence or all full-tilt clipping ear-bleeding overload.

The file size from different recorders may vary slightly, due to potentially different information recorded in the file header, but the audio portion of the WAV files from the different machines, running at the same recording rates for the exact same length of time, will be of identical size.

If data compressed, using either lossless or lossy compression, the file sizes will vary depending on the compression algorithm, the nature of the content and the similarity between channels.
« Last Edit: April 14, 2015, 12:03:16 PM by Gutbucket »
musical volition > vibrations > voltages > numeric values > voltages > vibrations> virtual teleportation time-machine experience
Better recording made easy - >>Improved PAS table<< | Made excellent- >>click here to download the Oddball Microphone Technique illustrated PDF booklet<< (note: This is a 1st draft, now several years old and in need of revision!  Stay tuned)

stevetoney

  • Guest
  • Trade Count: (0)
Re: file size hypothetical theoretical question
« Reply #3 on: April 14, 2015, 12:14:14 PM »
The file size from different recorders may vary slightly, due to potentially different information recorded in the file header, but the audio portion of the WAV files from the different machines, running at the same recording rates for the exact same length of time, will be of identical size.


If the clock speeds are slightly off on the two different recorders, that would alter the file size for the same duration of actual time since the number of samples would be different, no?  Over a minute, they'd still be very close, but not identical.
« Last Edit: April 14, 2015, 12:16:23 PM by tonedeaf »

Offline daspyknows

  • Complaint Dept.
  • Trade Count: (1)
  • Needs to get out more...
  • *****
  • Posts: 9636
  • Gender: Male
  • Don't ask, don't tell, don't get get caught
Re: file size hypothetical theoretical question
« Reply #4 on: April 14, 2015, 01:33:58 PM »
So no correlation in size of 2 flac recordings recorded with different gear and in this case size doesn't matter.   ;D

Offline easyed

  • Trade Count: (6)
  • Taperssection Member
  • ***
  • Posts: 277
  • get your Jam in the Can
Re: file size hypothetical theoretical question
« Reply #5 on: April 14, 2015, 01:51:02 PM »
So no correlation in size of 2 flac recordings recorded with different gear and in this case size doesn't matter.   ;D

consensus?

if 2 recordings were of different length, different bit depth or different sampling rates we would not expect the file sizes to be the same.

a recording of the same length made of silence or loud rock with the same recorder, same bit depth, same sampling rate will both be the same file size.

file size is irrelevant to sound quality.  absolutely no bearing.

not like, for example, microphone choice, placement, recorder choice, or transfer lineage, all of which can have a huge impact on sound quality.  or, for example, the difference between internal mics from the lawn at Shoreline and high end external mics in the sweet spot into a nice preamp and a nice recorder.
Beyerdynamic CK-930s > Naiant Tinybox or Littlebox > Sony PCM-M10 or
DPA 4061's > Core Sound Battery Box > Sony PCM-M10 or
matrix: Sound Devices 744T or
multitracking: Audient ASP008 preamps > JoeCo Blackbox BBR1B

Offline daspyknows

  • Complaint Dept.
  • Trade Count: (1)
  • Needs to get out more...
  • *****
  • Posts: 9636
  • Gender: Male
  • Don't ask, don't tell, don't get get caught
Re: file size hypothetical theoretical question
« Reply #6 on: April 14, 2015, 02:45:30 PM »
So no correlation in size of 2 flac recordings recorded with different gear and in this case size doesn't matter.   ;D

consensus?

if 2 recordings were of different length, different bit depth or different sampling rates we would not expect the file sizes to be the same.

a recording of the same length made of silence or loud rock with the same recorder, same bit depth, same sampling rate will both be the same file size.

file size is irrelevant to sound quality.  absolutely no bearing.

not like, for example, microphone choice, placement, recorder choice, or transfer lineage, all of which can have a huge impact on sound quality.  or, for example, the difference between internal mics from the lawn at Shoreline and high end external mics in the sweet spot into a nice preamp and a nice recorder.

True.  No different from the paradigm I deal with at work which is building Business Intelligence systems.  Often clients can't tie out their data because they load different inputs to 2 systems.  I often have to explain the only way they can tie out their two disparate systems with different data is clearing all the data since 0's always tie.  In other words, the file size will be the same before the recording starts 0 bytes.  Our "friend" Furburger will likely dispute this logic because he doesn't believe in science. 

Offline Gutbucket

  • record > listen > revise technique
  • Trade Count: (15)
  • Needs to get out more...
  • *****
  • Posts: 15698
  • Gender: Male
  • "Better to love music than respect it" ~Stravinsky
Re: file size hypothetical theoretical question
« Reply #7 on: April 14, 2015, 02:50:26 PM »
The file size from different recorders may vary slightly, due to potentially different information recorded in the file header, but the audio portion of the WAV files from the different machines, running at the same recording rates for the exact same length of time, will be of identical size.

If the clock speeds are slightly off on the two different recorders, that would alter the file size for the same duration of actual time since the number of samples would be different, no?  Over a minute, they'd still be very close, but not identical.

Yes, but that's covered by the "running at the same recording rates for the exact same length of time" part of the statement.  If the two clocks are not synced and are off slightly, then recording rates are not the same but slightly different, even if the machines are set to the same "nominal" sampling rate.  In other words, unless clock-synced both recorders aren't actually sampling at precisely 48kHz, although they will be very close to that rate.

if 2 recordings were of different length, different bit depth or different sampling rates we would not expect the file sizes to be the same.

a recording of the same length made of silence or loud rock with the same recorder, same bit depth, same sampling rate will both be the same file size.
^^
True.

Quote
file size is irrelevant to sound quality.  absolutely no bearing.

Not sure I follow this part.. If this is an extension of the the statement above it, then the file sizes should be identical.  In the real world case, two different files recorded on the same machine at the same rate and bit-depth will be have slightly different sizes if they were actually slightly different time durations. Similar to Tonedeaf's exception for two non-sync'd clocks, but in that case the time duration is presumed to be exactly the same, but the clocks are running at slightly different rates.
musical volition > vibrations > voltages > numeric values > voltages > vibrations> virtual teleportation time-machine experience
Better recording made easy - >>Improved PAS table<< | Made excellent- >>click here to download the Oddball Microphone Technique illustrated PDF booklet<< (note: This is a 1st draft, now several years old and in need of revision!  Stay tuned)

Offline easyed

  • Trade Count: (6)
  • Taperssection Member
  • ***
  • Posts: 277
  • get your Jam in the Can
Re: file size hypothetical theoretical question
« Reply #8 on: April 14, 2015, 06:26:05 PM »
The file size from different recorders may vary slightly, due to potentially different information recorded in the file header, but the audio portion of the WAV files from the different machines, running at the same recording rates for the exact same length of time, will be of identical size.

If the clock speeds are slightly off on the two different recorders, that would alter the file size for the same duration of actual time since the number of samples would be different, no?  Over a minute, they'd still be very close, but not identical.

Yes, but that's covered by the "running at the same recording rates for the exact same length of time" part of the statement.  If the two clocks are not synced and are off slightly, then recording rates are not the same but slightly different, even if the machines are set to the same "nominal" sampling rate.  In other words, unless clock-synced both recorders aren't actually sampling at precisely 48kHz, although they will be very close to that rate.

if 2 recordings were of different length, different bit depth or different sampling rates we would not expect the file sizes to be the same.

a recording of the same length made of silence or loud rock with the same recorder, same bit depth, same sampling rate will both be the same file size.
^^
True.

Quote
file size is irrelevant to sound quality.  absolutely no bearing.

Not sure I follow this part.. If this is an extension of the the statement above it, then the file sizes should be identical.  In the real world case, two different files recorded on the same machine at the same rate and bit-depth will be have slightly different sizes if they were actually slightly different time durations. Similar to Tonedeaf's exception for two non-sync'd clocks, but in that case the time duration is presumed to be exactly the same, but the clocks are running at slightly different rates.

the reason I mention quality, which of course is unrelated to file size, is that buttlicker's response to the posting of a better Fogerty recording than his was 'if it's better how comes is 80mb smaller?'
http://www.dimeadozen.org/torrents-details.php?id=524486&viewcomm=6821819#comm6821243
« Last Edit: April 14, 2015, 06:37:06 PM by easyed »
Beyerdynamic CK-930s > Naiant Tinybox or Littlebox > Sony PCM-M10 or
DPA 4061's > Core Sound Battery Box > Sony PCM-M10 or
matrix: Sound Devices 744T or
multitracking: Audient ASP008 preamps > JoeCo Blackbox BBR1B

Offline Gutbucket

  • record > listen > revise technique
  • Trade Count: (15)
  • Needs to get out more...
  • *****
  • Posts: 15698
  • Gender: Male
  • "Better to love music than respect it" ~Stravinsky
Re: file size hypothetical theoretical question
« Reply #9 on: April 14, 2015, 07:42:38 PM »
I don't currently have a Dime account so I can't see that linked page, however..

The simple answer in this case is that there is no direct correlation between files size and sound quality.  The files are slightly different sized empty containers which we can fill with whatever- good quality audio, crap audio, or with silence.  We can make a great sounding 16/44.1 WAV which smokes a 24/96 WAV, just by doing a better job of recording and putting much that better quality audio into that 3.5 times smaller container.  The good quality audio will fit within the smaller container easily.  The ultimate limits imposed by the bit depth and sample rate in this case is certainly not relevant (and is negligible in most cases).


But to address your actual question-
What exactly is 80MB smaller? the total file package?  Presumably these are FLAC data compressed files and the data compression will make the file sizes different, even if they are both 16bit/44.1kHz WAV files and happened to be exactly the same size uncompressed.  But the uncompressed WAVs are unlikely to be exactly the same size even if they are both 16/44.1 file sets, simply because the two file sets were tracked slightly differently by different folks, so each file and each entire set will not be exactly the same length.

In addition there could be extraneous data appended to the original WAVs, such as recorder manufacturer specific data.  For instance, file markers made on the R-44 fall into this category.   TLH removes that from the WAVs when FLAC compressing unless the Keep Foreign Metadata box is checked (I tell it not to strip that info when I compress my raw masters so I retain those markers in the FLAC'd but untracked master).

Beyond that, there could be different metadata in the two different FLAC sets, usually tagging info, but it could be anything, like photos.
musical volition > vibrations > voltages > numeric values > voltages > vibrations> virtual teleportation time-machine experience
Better recording made easy - >>Improved PAS table<< | Made excellent- >>click here to download the Oddball Microphone Technique illustrated PDF booklet<< (note: This is a 1st draft, now several years old and in need of revision!  Stay tuned)

Offline Gene Poole

  • Trade Count: (1)
  • Taperssection Regular
  • **
  • Posts: 104
Re: file size hypothetical theoretical question
« Reply #10 on: April 14, 2015, 07:47:53 PM »
the reason I mention quality, which of course is unrelated to file size, is that buttlicker's response to the posting of a better Fogerty recording than his was 'if it's better how comes is 80mb smaller?'
http://www.dimeadozen.org/torrents-details.php?id=524486&viewcomm=6821819#comm6821243

It's a meaningless assertion, but it could just be that the flac files were saved with a higher compression level.  Flac has 0-8 compression level settings.  the more compression, the more cpu needed to decompress, but not much of an issue on modern gear.

Offline jefflester

  • Trade Count: (2)
  • Taperssection All-Star
  • ****
  • Posts: 1560
  • Gender: Male
Re: file size hypothetical theoretical question
« Reply #11 on: April 14, 2015, 08:04:03 PM »
The biggest likely contributor to FLAC file size difference (for same show/~same length within seconds) is whether the original WAVs are normalized, compressed at all, or with peaks under (or well under) 0 dB. A fileset with low levels will compress to smaller FLAC than a fileset that has been normalized/compressed somewhat to make overall louder.
DPA4061 HEB -> R-09 / AT943 -> CA-UGLY -> R-09
AKG CK63 -> nBob actives -> Baby NBox -> R-09/DR2d
AKG CK63 -> AKG C460B -> Zoom F8/DR-680MKII
Line Audio CM4/Superlux S502/Samson C02/iSK Little Gem/Sennheiser E609/Shure SM57 -> Zoom F8/DR-680MKII (multitracked band recordings)

Offline Gutbucket

  • record > listen > revise technique
  • Trade Count: (15)
  • Needs to get out more...
  • *****
  • Posts: 15698
  • Gender: Male
  • "Better to love music than respect it" ~Stravinsky
Re: file size hypothetical theoretical question
« Reply #12 on: April 14, 2015, 09:19:25 PM »
I'm certainly no expert on the FLAC compression algorithm, but that makes sense.  Other audio attributes will also have an influence, such as how correlated the two channels are with each other (how similar or 'mono-ish', an X/Y recording will compress more than spaced omnis), and the nature of the sound itself (is it a recording of complex changing audio information, a simple and predictable sine wave, nothing but white noise, digital silence, or whatever, just to list some extremes).  Normalization making a large difference mentions makes sense because all the data space above peaking is entirely empty can be discarded.  By contrast the noise at the bottom is not empty and cannot be thrown discarded entirely.
musical volition > vibrations > voltages > numeric values > voltages > vibrations> virtual teleportation time-machine experience
Better recording made easy - >>Improved PAS table<< | Made excellent- >>click here to download the Oddball Microphone Technique illustrated PDF booklet<< (note: This is a 1st draft, now several years old and in need of revision!  Stay tuned)

Offline dabbler

  • Trade Count: (0)
  • Taperssection Member
  • ***
  • Posts: 359
Re: file size hypothetical theoretical question
« Reply #13 on: April 15, 2015, 04:37:56 AM »
FLAC definitely compresses quieter files smaller.

Find any unnormalized file and try encoding FLAC with sox, with and without normalization:

   # This command will normalize to -0.1 dBFS and encode to FLAC
   # the "stats" output will confirm the levels
   sox original.wav max.flac gain -n -0.1

   # Straight up encoding to FLAC only, again the "stats" output will just
   # confirm levels are unchanged.
   sox original.wav original.flac stats

Offline daspyknows

  • Complaint Dept.
  • Trade Count: (1)
  • Needs to get out more...
  • *****
  • Posts: 9636
  • Gender: Male
  • Don't ask, don't tell, don't get get caught
Re: file size hypothetical theoretical question
« Reply #14 on: April 15, 2015, 08:27:31 AM »
Furburger is too stupid to understand this.

"a "lossless" recording that is SMALLER.


man, you're rich, and I'm not talking in money or brain cells."


Offline jefflester

  • Trade Count: (2)
  • Taperssection All-Star
  • ****
  • Posts: 1560
  • Gender: Male
Re: file size hypothetical theoretical question
« Reply #15 on: April 15, 2015, 03:16:04 PM »
I did this simple test with a file I just happened to have handy. Not really representative of a typical live recording since it is a studio track, but nevertheless.

"Come Together" by The Beatles.
WAV is 43.7 MB
I take the WAV into Audition and drop the signal by 10 dB and save a new WAV, which is also 43.7 MB

01. Come Together - The Beatles -10 dB.WAV    43.7 MB
01. Come Together - The Beatles.WAV               43.7 MB

Then I FLAC them at various levels ( I did all Level 0 to 8, but these are representative enough)

FLAC0\01. Come Together - The Beatles -10 dB.flac'  (ratio = 0.520).       22.7 MB
FLAC0\01. Come Together - The Beatles.flac'             (ratio = 0.623).       27.2 MB

FLAC3\01. Come Together - The Beatles -10 dB.flac'  (ratio = 0.484).       21.1 MB
FLAC3\01. Come Together - The Beatles.flac'             (ratio = 0.586).       25.6 MB

FLAC8\01. Come Together - The Beatles -10 dB.flac'  (ratio = 0.471).       20.6 MB
FLAC8\01. Come Together - The Beatles.flac'             (ratio = 0.573).       25.0 MB

The content of the file had a bigger effect on the file size than the level of FLAC encoding. A 10 dB difference in overall RMS level made about a 20% difference in FLAC file size.

The ~10% difference (80MB) in size between daspyknows and furburgers Fogerty FLAC file sets could be accounted for probably ~5 dB difference in RMS level.  If someone wanted to compare the WAVs they would probably find something like that. I don't care enough to do so. I would venture to guess daspy's recording has more dynamic peaks and furburger's (from a cassette) has more compressed peaks.
DPA4061 HEB -> R-09 / AT943 -> CA-UGLY -> R-09
AKG CK63 -> nBob actives -> Baby NBox -> R-09/DR2d
AKG CK63 -> AKG C460B -> Zoom F8/DR-680MKII
Line Audio CM4/Superlux S502/Samson C02/iSK Little Gem/Sennheiser E609/Shure SM57 -> Zoom F8/DR-680MKII (multitracked band recordings)

Offline daspyknows

  • Complaint Dept.
  • Trade Count: (1)
  • Needs to get out more...
  • *****
  • Posts: 9636
  • Gender: Male
  • Don't ask, don't tell, don't get get caught
Re: file size hypothetical theoretical question
« Reply #16 on: April 16, 2015, 02:12:53 PM »
Here is what Furburger posted on my thread in response to this analysis


how retarded are you?

seriously?

you're now claiming that ***smaller*** files have "more dynamic peaks"?

seriously?

you don't know shit.

your show is inferior.

and it shows.

change your fucking board name already.

Offline Gutbucket

  • record > listen > revise technique
  • Trade Count: (15)
  • Needs to get out more...
  • *****
  • Posts: 15698
  • Gender: Male
  • "Better to love music than respect it" ~Stravinsky
Re: file size hypothetical theoretical question
« Reply #17 on: April 16, 2015, 02:20:33 PM »
seriously.
yes.
seriously.

[eye-roll]
musical volition > vibrations > voltages > numeric values > voltages > vibrations> virtual teleportation time-machine experience
Better recording made easy - >>Improved PAS table<< | Made excellent- >>click here to download the Oddball Microphone Technique illustrated PDF booklet<< (note: This is a 1st draft, now several years old and in need of revision!  Stay tuned)

Offline Sloan Simpson

  • Trade Count: (1)
  • Needs to get out more...
  • *****
  • Posts: 4013
  • Gender: Male
    • Southern Shelter
Re: file size hypothetical theoretical question
« Reply #18 on: April 16, 2015, 03:37:21 PM »
and it shows

Offline daspyknows

  • Complaint Dept.
  • Trade Count: (1)
  • Needs to get out more...
  • *****
  • Posts: 9636
  • Gender: Male
  • Don't ask, don't tell, don't get get caught
Re: file size hypothetical theoretical question
« Reply #19 on: April 16, 2015, 04:08:11 PM »
Dude's been drinking too much of this  :facepalm:

Offline daspyknows

  • Complaint Dept.
  • Trade Count: (1)
  • Needs to get out more...
  • *****
  • Posts: 9636
  • Gender: Male
  • Don't ask, don't tell, don't get get caught
Re: file size hypothetical theoretical question
« Reply #20 on: April 19, 2015, 01:42:14 PM »
I recorded a show last night and as part of my processing the files for upload I ran a test.  Same fileset, but 2 copies

Original Wav files 1.25GB

First fileset FLAC level 8  (best per TLH) 649 MB
Second fileset FLAC level 0 (fastest per TLH) 704MB

Wav files compressed using Winzip 1.07GB

that is a 7.4% reduction in size attributable only to what level of FLAC compression is used.   

 

RSS | Mobile
Page created in 0.089 seconds with 49 queries.
© 2002-2024 Taperssection.com
Powered by SMF