ATRAC Advanced Lossless decoding, not bit-exact?

[macdavis]

Summary of the bug:
I converted "dts_sample.flac" to ATRAC Advanced Lossless ("dts_sample.oma") with SonicStage (Version 4.3, download link ​ftp://ftp.vaio-link.com/pub/Downloads/SO/A/SST/SOASST-01198207-UN.exe). Then, using ffmpeg to convert "dts_sample.oma" back to Flac. However, the conversion seems not bit-exact.

I checked the original Flac with Adobe Audition. Here is the original one's amplitude statistics.

	Channel 1	Channel 2
Peak Amplitude:	-0.09 dB	-1.08 dB
True Peak Amplitude:	-0.09 dBTP	-1.08 dBTP
Maximum Sample Value:	25664	27834
Minimum Sample Value:	-32440	-28930
Possibly Clipped Samples:	0	0
Total RMS Amplitude:	-24.68 dB	-24.34 dB
Maximum RMS Amplitude:	-8.98 dB	-7.12 dB
Minimum RMS Amplitude:	-84.55 dB	-83.76 dB
Average RMS Amplitude:	-38.58 dB	-38.89 dB
DC Offset:	0.05 %	0.02 %
Measured Bit Depth:	16	16
Dynamic Range:	75.57 dB	76.65 dB
Dynamic Range Used:	73.20 dB	74.15 dB
Loudness (Legacy):	-12.45 dB	-11.96 dB
Perceived Loudness (Legacy):	-15.06 dB	-11.62 dB
ITU-R BS.1770-3 Loudness: -19.83 LUFS

0dB = FS Square Wave
Using RMS Window of 50.00 ms
Account for DC = true

Here is the amplitude statistics of Flac converted from "dts_sample.oma"

	Channel 1	Channel 2
Peak Amplitude:	-0.03 dB	-1.13 dB
True Peak Amplitude:	-0.03 dBTP	-1.13 dBTP
Maximum Sample Value:	25292.57	28165.01
Minimum Sample Value:	-32656.77	-28762.40
Possibly Clipped Samples:	0	0
Total RMS Amplitude:	-24.71 dB	-24.38 dB
Maximum RMS Amplitude:	-9.00 dB	-7.22 dB
Minimum RMS Amplitude:	-84.51 dB	-84.95 dB
Average RMS Amplitude:	-38.73 dB	-39.15 dB
DC Offset:	0.05 %	0.01 %
Measured Bit Depth:	24	24
Dynamic Range:	75.51 dB	77.72 dB
Dynamic Range Used:	73.60 dB	75.05 dB
Loudness (Legacy):	-13.47 dB	-11.95 dB
Perceived Loudness (Legacy):	-18.84 dB	-11.55 dB
ITU-R BS.1770-3 Loudness: -19.84 LUFS

0dB = FS Square Wave
Using RMS Window of 50.00 ms
Account for DC = true

How to reproduce:

ffmpeg -i /Volumes/New\ Volume/sony\ lossless/dts_sample.oma -c:a flac -compression_level 12 /Volumes/New\ Volume/sony\ lossless/dts_sample_oma_to_flac.flac 
ffmpeg version 3.3.git Copyright (c) 2000-2017 the FFmpeg developers
  built with Apple LLVM version 8.1.0 (clang-802.0.42)
  configuration: --optflags=-Ofast --prefix=/usr/local --extra-cflags='-march=haswell -mtune=intel -m64 -minline-all-stringops -fomit-frame-pointer -fforce-addr -finline-functions -funit-at-a-time -funroll-loops -ffunction-sections -fdata-sections -Ofast -pipe -fomit-frame-pointer -ftree-vectorize' --target-os=darwin --arch=x86_64 --cpu=haswell --disable-runtime-cpudetect --disable-debug --disable-opencl --disable-avdevice --disable-ffplay --disable-ffprobe --disable-postproc --disable-ffserver --disable-devices --disable-avresample --disable-indev=qtkit --disable-doc --enable-pthreads --enable-version3 --enable-hardcoded-tables --enable-gpl --enable-nonfree
  libavutil      55. 67.100 / 55. 67.100
  libavcodec     57.100.103 / 57.100.103
  libavformat    57. 75.100 / 57. 75.100
  libavfilter     6. 95.100 /  6. 95.100
  libswscale      4.  7.101 /  4.  7.101
  libswresample   2.  8.100 /  2.  8.100
Input #0, oma, from '/Volumes/New Volume/sony lossless/dts_sample.oma':
  Metadata:
    title           : dts_sample
    OMG_TRACK       : 0
    OMG_TIT2S       : dts_sample
    TLEN            : 18894
  Duration: N/A, start: 0.000000, bitrate: N/A
    Stream #0:0: Audio: atrac3pal ([33][0][0][0] / 0x0021), 44100 Hz, stereo, fltp
Stream mapping:
  Stream #0:0 -> #0:0 (atrac3pal (atrac3plusal) -> flac (native))
Press [q] to stop, [?] for help
[flac @ 0x7ffb1a801800] encoding as 24 bits-per-sample
Output #0, flac, to '/Volumes/New Volume/sony lossless/dts_sample_oma_to_flac.flac':
  Metadata:
    title           : dts_sample
    OMG_TRACK       : 0
    OMG_TIT2S       : dts_sample
    TLEN            : 18894
    encoder         : Lavf57.75.100
    Stream #0:0: Audio: flac, 44100 Hz, stereo, s32 (24 bit), 128 kb/s
    Metadata:
      encoder         : Lavc57.100.103 flac
size=    2711kB time=00:00:18.94 bitrate=1172.2kbits/s speed=20.1x    
video:0kB audio:2703kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.302080%

[Carl Eugen Hoyos]

Does the SonicStage encoder support float-in-wav input as produced with ffmpeg -i input -acodec pcm_f32le out.wav?

[macdavis]

Replying to cehoyos:

Does the SonicStage encoder support float-in-wav input as produced with ffmpeg -i input -acodec pcm_f32le out.wav?

No, it doesn't. It will show the following error message.

The following files cannot be converted to ATRAC Advanced Lossless.
Only WAV (16bit 44.1kHz 2ch) tracks can be compressed to ATRAC Advanced Lossless format.

[Elon Musk]

Duplicate of #5334.