webbie64
Headphoneus Supremus
- Joined
- Oct 23, 2006
- Posts
- 1,642
- Likes
- 12
So many threads ask and analyse this question: "Can YOU hear the difference between various lossy compressed, lossless compressed and uncompressed formats?"
Engaging discussions have emerged in these previous threads and I appreciate all that I have gained from the various contributors.
The conclusions I gain from these more active, debated threads, though, is that I really have to take everything said with a significant dose of many grains of salt and simply 'bite the bullet', invest the time, and reach my own conclusions: for MY ears and MY equipment.
So, here goes...
Background: I was looking at the question mainly for my portable and transportable rigs, and for the purpose of archiving.
Equipment: Sony NW-HD5 or iRiver H140/120 (all with a very good quality line out) fed via a can'tsleep pure silver mini-2-mini into a well burned in RSA Hornet to Westone UM2s, OR
iRiver H140/120 (both with a very good quality optical out) fed via a Van Den Hul Optocoupler MKII mini-mini toslink cable to an AOS Piccolo DAC sent via VH Cryo Pulsars to a Stax SRD-XPro driving Sennheiser HE60s, Stax SR-001 MkIIs, Stax Lambdas or Stax Sigmas OR
iRiver H140/120 (both with a very good quality optical out) fed via a Van Den Hul Optocoupler MKII mini-mini toslink cable to a Optical/SPDIF converter to an Apuresound DAC-1 thence via Moon Audio Silver Dragons to either McAllister EA-4 or Woo Audio GES (both unmodded) to Sennheiser HE60s, Stax SR-001 MkIIs, Stax Lambdas and Stax Sigmas.
Note: The UM2 terminated rig is the portable rig which, in its lowest profile form is MP3 based (Sony DAP). The other are home-based rigs (Bed/transportable and Study). The iRiver sources can also be used with the home speaker-based rig but I heard sufficient consistent differences through the headphone-terminated rigs to post these comments.
The encoding: WAV, FLAC (Level 5 and Level 8), MP3 (320 CBR, 32-320 VBR, 224-320 ABR) all encoded via EAC. MP3 compressed via LAME 3.97.
The conclusions (i.e. MY take):
I come from a background that has included many forms of home analogue (vinyl, reel-to-reel tape, audio cassettes, etc) as well as the later digital sources. I therefore find that the available sizes of HDDs are more than sufficient to cope with the tracks from my collection that I want access to everywhere and anytime. As a result I am looking strictly at SQ. I want the absolute best I can get from my equipment and therefore want to feed at the absolute front end (i.e. the source files) the best sounding data.
So, from a pure SQ POV:
WAV is the standard. Well extracted via EAC the WAV provides the same audio data as the CD source. It sounds the same in all terms - Acoustical absorption, Audibility (of flaws), Detail, Distance, Frequency response - ENTIRE RANGE/BASS(Amount/Smoothness/Extension)/MIDRANGE/LOW & MID TREBLE/EXTREME TREBLE, Imaging, Noise, Realism, Soundstaging (WIDTH/CONTINUITY/DEPTH), Texture, Timbre, Transient response.
FLAC supports what others have written about the Lossless compressor - it unpacks identically to the WAV source file. As a result it is as good as WAV in SQ. I do NOT find it in any way brighter or changed from the original. This applies to both Level 5 and Level 8 compression. In terms of compression of choice, therefore, I do not have SQ as a determining factor but, rather, what has already been discussed elsewhere - Level 5 encodes and decodes faster, with less processing power VERSUS saved space from the extra compression allows extra tracks per HDD. I tend to agree with those who advocate Level 5. Although I agree the overall encoding time is unchanged if you encode to FLAC at the same time as extracting the next WAV, I find that the size savings are minimal for the amount of extra decoding processing power utilised. I have not done a solid comparison over an extended time on this (remember my review is all about SQ) but the extra 2-4MB saved per track provides me with one extra track per 5 or so tracks on a HDD that is not particularly full AND the battery power will never fully play the whole HDD regardless but will produce a similar reduction in battery resources over time (meaning potentially up to 1/5 less time per battery charge). I'll take the extra battery time over the extra available tracks. Your choice might be different to mine.
MP3. Aaah, MP3. Like VHS video cassettes I marvel at the way we as an overall human society embrace a technology that is less than the best because it's more accessible at a slightly earlier time. First to the mass market really does count .
I only look at this compressed format because it's usable across all of my equipment - all car & home players - plus is most readily accepted should I wish to share a track with a family member or friend. And, yes, I was an early adopter and have stuck with the format.
The three variations of compression have been chosen for solid reasons: 320CBR is, of course, considered the highest quality common format (although higher kbps are now possible, most players top out at 320 kbps). This is a simple compression yet potentially wasteful of HDD space for the level of data required. 32-320 VBR is meant to address the potential HDD wastage by encoding at the best compression level for each segment of data. Most of my players support it (one car player doesn't). 224-320 ABR is what I have used prior to this comparison as it seemed good enough and played on all of my players (including the car player that won't handle full VBR).
The conclusion, for me, is that the other two formats do NOT match 320 kbps CBR. Yes, I CAN hear differences - across multiple tracks on multiple rigs. Yes, the differences are consistent across the rigs (although the more revealing the rig, the more apparent the difference).
In comparison to 320 kbps I find voices to be slightly rougher with the other two variations. This is most apparent when comparing CBR to VBR (VBR definitely has a roughness - perhaps because it's choosing over such a wider range of bitrates? - ABR is definitely closer to CBR and my choice has been a much smaller bitrate range). In the mid-highs area the VBR and, to some extent, ABR variants are slightly cloudier, less defining of instrument/voice placement and, as a consequence, depth and soundstage. Not that any of them can get close to the depth and soundstage of the lossless or uncompressed formats. For instance, if a WAV/FLAC gets a rating of '10', then the MP3 compressions are rated 320 CBR = '7.50', 224-320 ABR = '7.40', 32-320 VBR = '7.15'.
In terms of SQ differences amongst the MP3s I believe most listeners would not be as finicky as me. Most could settle into the quite reasonable ABR and VBR versions for the considerable space saved (and therefore greater range of songs accessible on the relevant HDD). The differences ARE SMALL. They are much less than the differences between uncompressed/lossless and 320 kbps CBR. And, yes, for most tracks I might easily accept the small additional SQ compromise (of ABR) considering the compromise already accepted in listening to the MP3 version over the Lossless version.
But, in terms of strict A-B comparison, 320 kbps CBR is the better sounding and, for me, that difference is both noticeable and appreciated and, therefore, to me, worth it.
So my HDDs are being gradually reloaded with FLAC Level 5 and MP3 320 kbps CBR. Yes, the MP3 player will be much fuller with potentially less songs but, yes, for those who ask the question, the isolation afforded by my portable rig does make the difference worth it for me. (And this may well improve when I get some Livewires
).
Obviously my overarching drive towards SQ is what drives my choices here. I've got the HDDs and HDD space to spare for the volume of collection I wish to transport with me. Your choices, and compromises, may well be different.
Thanks for reading. Hope this helps others. Happy Head-Fi-ing!
Engaging discussions have emerged in these previous threads and I appreciate all that I have gained from the various contributors.
The conclusions I gain from these more active, debated threads, though, is that I really have to take everything said with a significant dose of many grains of salt and simply 'bite the bullet', invest the time, and reach my own conclusions: for MY ears and MY equipment.
So, here goes...
Background: I was looking at the question mainly for my portable and transportable rigs, and for the purpose of archiving.
Equipment: Sony NW-HD5 or iRiver H140/120 (all with a very good quality line out) fed via a can'tsleep pure silver mini-2-mini into a well burned in RSA Hornet to Westone UM2s, OR
iRiver H140/120 (both with a very good quality optical out) fed via a Van Den Hul Optocoupler MKII mini-mini toslink cable to an AOS Piccolo DAC sent via VH Cryo Pulsars to a Stax SRD-XPro driving Sennheiser HE60s, Stax SR-001 MkIIs, Stax Lambdas or Stax Sigmas OR
iRiver H140/120 (both with a very good quality optical out) fed via a Van Den Hul Optocoupler MKII mini-mini toslink cable to a Optical/SPDIF converter to an Apuresound DAC-1 thence via Moon Audio Silver Dragons to either McAllister EA-4 or Woo Audio GES (both unmodded) to Sennheiser HE60s, Stax SR-001 MkIIs, Stax Lambdas and Stax Sigmas.
Note: The UM2 terminated rig is the portable rig which, in its lowest profile form is MP3 based (Sony DAP). The other are home-based rigs (Bed/transportable and Study). The iRiver sources can also be used with the home speaker-based rig but I heard sufficient consistent differences through the headphone-terminated rigs to post these comments.
The encoding: WAV, FLAC (Level 5 and Level 8), MP3 (320 CBR, 32-320 VBR, 224-320 ABR) all encoded via EAC. MP3 compressed via LAME 3.97.
The conclusions (i.e. MY take):
I come from a background that has included many forms of home analogue (vinyl, reel-to-reel tape, audio cassettes, etc) as well as the later digital sources. I therefore find that the available sizes of HDDs are more than sufficient to cope with the tracks from my collection that I want access to everywhere and anytime. As a result I am looking strictly at SQ. I want the absolute best I can get from my equipment and therefore want to feed at the absolute front end (i.e. the source files) the best sounding data.
So, from a pure SQ POV:
WAV is the standard. Well extracted via EAC the WAV provides the same audio data as the CD source. It sounds the same in all terms - Acoustical absorption, Audibility (of flaws), Detail, Distance, Frequency response - ENTIRE RANGE/BASS(Amount/Smoothness/Extension)/MIDRANGE/LOW & MID TREBLE/EXTREME TREBLE, Imaging, Noise, Realism, Soundstaging (WIDTH/CONTINUITY/DEPTH), Texture, Timbre, Transient response.
FLAC supports what others have written about the Lossless compressor - it unpacks identically to the WAV source file. As a result it is as good as WAV in SQ. I do NOT find it in any way brighter or changed from the original. This applies to both Level 5 and Level 8 compression. In terms of compression of choice, therefore, I do not have SQ as a determining factor but, rather, what has already been discussed elsewhere - Level 5 encodes and decodes faster, with less processing power VERSUS saved space from the extra compression allows extra tracks per HDD. I tend to agree with those who advocate Level 5. Although I agree the overall encoding time is unchanged if you encode to FLAC at the same time as extracting the next WAV, I find that the size savings are minimal for the amount of extra decoding processing power utilised. I have not done a solid comparison over an extended time on this (remember my review is all about SQ) but the extra 2-4MB saved per track provides me with one extra track per 5 or so tracks on a HDD that is not particularly full AND the battery power will never fully play the whole HDD regardless but will produce a similar reduction in battery resources over time (meaning potentially up to 1/5 less time per battery charge). I'll take the extra battery time over the extra available tracks. Your choice might be different to mine.
MP3. Aaah, MP3. Like VHS video cassettes I marvel at the way we as an overall human society embrace a technology that is less than the best because it's more accessible at a slightly earlier time. First to the mass market really does count .
I only look at this compressed format because it's usable across all of my equipment - all car & home players - plus is most readily accepted should I wish to share a track with a family member or friend. And, yes, I was an early adopter and have stuck with the format.
The three variations of compression have been chosen for solid reasons: 320CBR is, of course, considered the highest quality common format (although higher kbps are now possible, most players top out at 320 kbps). This is a simple compression yet potentially wasteful of HDD space for the level of data required. 32-320 VBR is meant to address the potential HDD wastage by encoding at the best compression level for each segment of data. Most of my players support it (one car player doesn't). 224-320 ABR is what I have used prior to this comparison as it seemed good enough and played on all of my players (including the car player that won't handle full VBR).
The conclusion, for me, is that the other two formats do NOT match 320 kbps CBR. Yes, I CAN hear differences - across multiple tracks on multiple rigs. Yes, the differences are consistent across the rigs (although the more revealing the rig, the more apparent the difference).
In comparison to 320 kbps I find voices to be slightly rougher with the other two variations. This is most apparent when comparing CBR to VBR (VBR definitely has a roughness - perhaps because it's choosing over such a wider range of bitrates? - ABR is definitely closer to CBR and my choice has been a much smaller bitrate range). In the mid-highs area the VBR and, to some extent, ABR variants are slightly cloudier, less defining of instrument/voice placement and, as a consequence, depth and soundstage. Not that any of them can get close to the depth and soundstage of the lossless or uncompressed formats. For instance, if a WAV/FLAC gets a rating of '10', then the MP3 compressions are rated 320 CBR = '7.50', 224-320 ABR = '7.40', 32-320 VBR = '7.15'.
In terms of SQ differences amongst the MP3s I believe most listeners would not be as finicky as me. Most could settle into the quite reasonable ABR and VBR versions for the considerable space saved (and therefore greater range of songs accessible on the relevant HDD). The differences ARE SMALL. They are much less than the differences between uncompressed/lossless and 320 kbps CBR. And, yes, for most tracks I might easily accept the small additional SQ compromise (of ABR) considering the compromise already accepted in listening to the MP3 version over the Lossless version.
But, in terms of strict A-B comparison, 320 kbps CBR is the better sounding and, for me, that difference is both noticeable and appreciated and, therefore, to me, worth it.
So my HDDs are being gradually reloaded with FLAC Level 5 and MP3 320 kbps CBR. Yes, the MP3 player will be much fuller with potentially less songs but, yes, for those who ask the question, the isolation afforded by my portable rig does make the difference worth it for me. (And this may well improve when I get some Livewires
Obviously my overarching drive towards SQ is what drives my choices here. I've got the HDDs and HDD space to spare for the volume of collection I wish to transport with me. Your choices, and compromises, may well be different.
Thanks for reading. Hope this helps others. Happy Head-Fi-ing!