FLAC sound on Sansa Clip+

[Yard-Ape]

Greetings,

After some fatigue a minute into a FLAC track ("Snake As Road Sign" by Evan Parker on the album "Atlanta"), I experimented, and discovered that the Sansa stock firmware plays the track with much greater clarity and much less upper-mid distortion than Rockbox does.  The staging and lower end resonance is also superior with the stock firmware.  I flattened out the EQ and defaulted all the other settings I could find.  I tested with and without amplification.  I'm using Rockbox 3.9.

I was surprised at this, and I expect some incredulity here, but I trust my ears on this occasion at least (I've been a musician for 30 years and my hearing is at least not very poor).  All things being equal I can't normally tell the difference between FLAC and 320kbps Lame mp3 files with the kind of hardware I use; anything lower than 192kbps usually hurts a bit.  But the difference in this case between the two FLAC implementations was so great as to be undeniable.  The particular track I was playing is very demanding in the upper-mid range and I'm sure any of you would have been able to hear it.

Should this be expected?  Is there anything I can do to remedy it?

(PS. I was particularly surprised that this could be a software issue given the fact that the stock firmware eats through battery life very quickly when playing FLAC.  I take that to mean that Sandisk's implementation of the decoder is at least not very efficient.  Maybe the software difference is further upstream than the decoder?)

Thanks in advance!

[saratoga]

The Rockbox flac decoder is lossless. If it sounds different then you have enabled some DSP effects or are comparing it to something with DSP effects.

[Yard-Ape]

Okay.  Thank you.

[amayer]

I'm using Rockbox 3.9.

I dont have a sansa clip+ and im not sure when the last time the FLAC codec was updated but i would recomend upgrading to version 3.11 and seeing if that makes a difference.

[Julian67]

I'm not sure this is relevant but it might be:

What isn't obvious on a first install or without exploring the settings and reading the manual is that the setting "Track Gain If Shuffling" is on by default, and would perhaps be better described as "Replay Gain Album Mode On Unless Shuffling".  This means the new or uninformed user may think they are running RB with no processing but in fact they are running RG with no pre-amp modifier or clipping protection.  There can sometimes be a very significant volume boost without any clipping protection.   I wonder if some of the recent posts about unexpected audio quality arise from increased levels without clipping protection.  The typical expected use case of RG seems to be for recordings that need attenuation (much popular music) and which have negative album RG values,  but lots of instrumental and vocal recordings (much non-orchestral classical and pre-classical music) have surprisingly high positive album RG values (+4 dB is not unusual and I think somewhere I even have an album that is +8 or +9).

I recall that I didn't realise that RockBox ran with RG by default until I experienced several tracks clipping, which prompted me to investigate and to Read The Fine Manual.

[JohnP]

Replay-gain issues aside :
The stock clip-firmware and rockbox do not sound the same,
stock firmware has the pitch-error, rockbox nearly doesn't .
A440Hz (middle A) is actually A441Hz on a clip with original firmware.

This is also known as 'pitch-inflation' and it has real audible effects :

"During historical periods when instrumental music rose in prominence (relative to the voice), there was a continuous tendency for pitch levels to rise. This "pitch inflation" seemed largely a product of instrumentalists' competing with each other, each attempting to produce a brighter, more "brilliant", sound than that of their rivals.
(In string instruments, this is not all acoustic illusion: when tuned up, they actually sound objectively brighter because the higher string tension results in larger amplitudes for the harmonics.)
This tendency was also prevalent with wind instrument manufacturers, who crafted their instruments to play generally at a higher pitch than those made by the same craftsmen years earlier."

https://en.wikipedia.org/wiki/History_of_pitch_standards_in_Western_music#History_of_pitch_standards_in_Western_music

[Julian67]

A pitch error in player firmware is definitely not the same thing as a historical phenomenon of instrument makers and players changing manufacturing processes and playing styles to achieve higher levels and/or higher pitch.

It may that the pitch error in the Sansa firmware accounts for the fault Yard-Ape perceives to be associated with playback in Rockbox because typically people would not start by assuming that Sansa ship players with a perceptible pitch error.  If that's the case then there isn't an issue with playback in Rockbox, but Yard-Ape may have become so familiar with hearing some music off pitch that it now sounds odd when played back correctly.  Ouch.

[Julian67]

I was curious about this audio issue so I decided to get hold of the album for the fair use purpose of trying to detect and/or diagnose the supposed playback fault.  I discovered it is long out of print and impossible to buy here in the UK.  Finally I was able to download a lossless copy, (ape+cue with a log suggesting a CD source and an error free rip).  I split the lossless image into separate tracks, encoded to Ogg Vorbis, and applied Replay Gain.  The album gain is 1.38 dB but the track gain for the track "The Snake As Road Sign" is much higher 4.32 dB.  The music itself is violently dissonant (that is a description and not a judgement) saxophone solo.  I first played it in mplayer on my desktop and after the initial shock transferred it to my Sansa Clip+ which dual boots original firmware and RockBox 3.11.2.

I can only say that I can find no difference in the audio playback between Rockbox, Sansa Original Firmware, or playback on my desktop via mplayer,  except in output level.  This music doesn't have any complexity (it's clearly unusual music but that is a  different matter), its dynamic range is rather narrow, it is in no way a loud master and it presents no playback challenges or issues of any kind.  I can only wonder if Yard-Ape's initial post is indeed factual and not intended as an exercise in jazz hipster irony.

Code: [Select]

$ mediainfo 03\ -\ The\ Snake\ As\ Road\ Sign.ogg 
General
Complete name                            : 03 - The Snake As Road Sign.ogg
Format                                   : OGG
File size                                : 22.6 MiB
Duration                                 : 17mn 6s
Overall bit rate mode                    : Variable
Overall bit rate                         : 185 Kbps
Album replay gain                        : 1.38 dB
Album replay gain peak                   : 0.906816
Album                                    : Atlanta
Track name                               : The Snake As Road Sign
Track name/Position                      : 03
Performer                                : Evan Parker
Genre                                    : hipster jazz dissonance (I invented this genre a minute ago)
Recorded date                            : 1990

Audio
ID                                       : 964310449 (0x397A35B1)
Format                                   : Vorbis
Format settings, Floor                   : 1
Duration                                 : 17mn 6s
Bit rate mode                            : Variable
Bit rate                                 : 224 Kbps
Channel(s)                               : 2 channels
Sampling rate                            : 44.1 KHz
Compression mode                         : Lossy
Replay gain                              : 4.32 dB
Replay gain peak                         : 0.327418
Stream size                              : 27.4 MiB
Writing library                          : libVorbis (Everywhere) (20100325 (Everywhere))