For people complaining processing and file size, bear in mind that uncompressed audio is literally x500 smaller than uncompressed video and it has been around for 20 years in blu ray disc.
Audio Fake Detector PRO
- Thread starter Alessandro Comito
- Start date
- Status
The analysis takes roughly the same time on a 1 MB file as on a 100 MB (or even larger) file because the program does not analyze the entire file.For people complaining processing and file size, bear in mind that uncompressed audio is literally x500 smaller than uncompressed video and it has been around for 20 years in blu ray disc.
Instead of processing the whole track second by second, the tool only takes 4-5 short "snapshots" (called slots) at specific points in the song.
You can see it clearly in this example: slot@98s → slot@176s → slot@294s → slot@372s → end@442s
Only these short segments are analyzed with the Hi-Res spectrogram (even though the track is 7 minutes and 22 seconds long).
This applies both to the length of the track (2 minutes or 2 hours makes very little difference) and to the sample rate in kHz (44.1 kHz, 96 kHz, 192 kHz).
A 192 kHz file is much larger because it contains about 4× more samples per second than a 48 kHz file.
However, since the tool only analyzes a few brief segments instead of the entire data stream, the processing time remains almost identical.
In short:
- It is not decoding and analyzing hundreds of megabytes of audio.
- It is only looking at a few scattered seconds of the track, regardless of resolution (kHz) or total duration.
- So it's not thanks to a super-fast CPU, but to the smart, efficient method the tool uses (slot-based sampling).
Result: nearly constant analysis time, whether the file is small or a huge high-resolution one.
P.S. Version 7.7 was released on 2026-06-22.
Last edited:
I have down and up converted my vinyl 24/192 rips up to 384khz and down to 96khz to see the results and this is what I got .
[15:09:25] auCDtect : C:\Users\Vladimir\AppData\Local\AudioFakeDetector\auCDtect.exe
[15:09:25] ffmpeg : C:\Users\Vladimir\AppData\Local\AudioFakeDetector\ffmpeg.exe
[15:09:25] ffprobe : C:\Users\Vladimir\AppData\Local\AudioFakeDetector\ffprobe.exe
[15:09:25] ffmpeg update check skipped (last checked today at 15:07).
[15:09:25] All dependencies OK.
[15:09:25]
[15:09:25] Found 3 audio file(s) to analyse (including hidden).
[15:09:25]
[15:09:26] [1/3] C:\Users\Vladimir\Desktop\12. Darude - Sandstorm96.flac
[15:09:26] Full analysis (>= 170 kbps) │ Engine: spectrogram (Hi-Res 96000 Hz - auCDtect bypassed)
[15:09:26] No encoder tag │ flac │ 96 kHz │ 2ch │ 24 bit │ 3152 kbps │ 00:07:23
[15:09:31] slot@89s(rms=0.009) -> cutoff=44762 Hz | wall=False | ratio=0.843 | peak=79.3 | NR=False
[15:09:32] slot@157s(rms=0.151) -> cutoff=46311 Hz | wall=False | ratio=0.224 | peak=99.9 | NR=False
[15:09:33] random@251s(rms=0.078) -> cutoff=45654 Hz | wall=False | ratio=0.487 | peak=88.9 | NR=False
[15:09:34] slot@374s(rms=0.18) -> cutoff=46358 Hz | wall=False | ratio=0.233 | peak=97.8 | NR=False
[15:09:36] end@420s -> cutoff=45889 Hz | wall=False | ratio=0.412 | peak=80.7 | NR=False
[15:09:36] [OK] Spectrogram Hi-Res: genuine content | cutoff ~46358 Hz (> 28 kHz)
[15:09:36]
[15:09:36] [2/3] C:\Users\Vladimir\Desktop\12. Darude - Sandstorm384.flac
[15:09:37] Full analysis (>= 170 kbps) │ Engine: spectrogram (Hi-Res 384000 Hz - auCDtect bypassed)
[15:09:37] No encoder tag │ flac │ 384 kHz │ 2ch │ 24 bit │ 7652 kbps │ 00:07:23
[15:09:47] slot@89s(rms=0.009) -> cutoff=92340 Hz | wall=False | ratio=0.415 | peak=81.5 | NR=False
[15:09:49] slot@157s(rms=0.151) -> cutoff=91965 Hz | wall=False | ratio=0.586 | peak=133.3 | NR=False
[15:09:50] random@251s(rms=0.078) -> cutoff=92152 Hz | wall=False | ratio=0.502 | peak=118.1 | NR=False
[15:09:52] slot@374s(rms=0.18) -> cutoff=91965 Hz | wall=False | ratio=0.59 | peak=126.4 | NR=False
[15:09:53] end@420s -> cutoff=92340 Hz | wall=False | ratio=0.421 | peak=87.6 | NR=False
[15:09:54] [OK] Spectrogram Hi-Res: genuine content | cutoff ~92340 Hz (> 28 kHz)
[15:09:54]
[15:09:54] [3/3] C:\Users\Vladimir\Downloads\Trance Classics HD Vinyl Rips\12. Darude - Sandstorm.flac
[15:09:54] Full analysis (>= 170 kbps) │ Engine: spectrogram (Hi-Res 192000 Hz - auCDtect bypassed)
[15:09:54] Encoder: fre:ac v1.1.6 (20230210) │ flac │ 192 kHz │ 2ch │ 24 bit │ 5639 kbps │ 00:07:23
[15:10:12] slot@89s(rms=0.009) -> cutoff=96000 Hz | wall=False | ratio=0 | peak=79.8 | NR=False
[15:10:13] slot@157s(rms=0.151) -> cutoff=96000 Hz | wall=False | ratio=0 | peak=110.7 | NR=False
[15:10:14] random@251s(rms=0.078) -> cutoff=96000 Hz | wall=False | ratio=0 | peak=98.8 | NR=False
[15:10:15] slot@374s(rms=0.18) -> cutoff=96000 Hz | wall=False | ratio=0 | peak=106 | NR=False
[15:10:17] end@420s -> cutoff=96000 Hz | wall=False | ratio=0 | peak=82 | NR=False
[15:10:18] [OK] Spectrogram Hi-Res: genuine content | cutoff ~96000 Hz (> 28 kHz)
[15:10:18]
[15:10:18]
[15:10:18] =========================================================================================================
[15:10:18] SUMMARY
[15:10:18] =========================================================================================================
[15:10:18] Files found : 3 (including hidden)
[15:10:18] Analyzed : 3
[15:10:18] Skipped : 0
[15:10:18] OK : 3
[15:10:18] FAKE : 0
[15:10:18] SUSPECT : 0
[15:10:18] UNKNOWN : 0
[15:10:18] Orphan dirs : N/A (file-mode scan)
[15:10:18] ===================================================================================================
[15:09:25] auCDtect : C:\Users\Vladimir\AppData\Local\AudioFakeDetector\auCDtect.exe
[15:09:25] ffmpeg : C:\Users\Vladimir\AppData\Local\AudioFakeDetector\ffmpeg.exe
[15:09:25] ffprobe : C:\Users\Vladimir\AppData\Local\AudioFakeDetector\ffprobe.exe
[15:09:25] ffmpeg update check skipped (last checked today at 15:07).
[15:09:25] All dependencies OK.
[15:09:25]
[15:09:25] Found 3 audio file(s) to analyse (including hidden).
[15:09:25]
[15:09:26] [1/3] C:\Users\Vladimir\Desktop\12. Darude - Sandstorm96.flac
[15:09:26] Full analysis (>= 170 kbps) │ Engine: spectrogram (Hi-Res 96000 Hz - auCDtect bypassed)
[15:09:26] No encoder tag │ flac │ 96 kHz │ 2ch │ 24 bit │ 3152 kbps │ 00:07:23
[15:09:31] slot@89s(rms=0.009) -> cutoff=44762 Hz | wall=False | ratio=0.843 | peak=79.3 | NR=False
[15:09:32] slot@157s(rms=0.151) -> cutoff=46311 Hz | wall=False | ratio=0.224 | peak=99.9 | NR=False
[15:09:33] random@251s(rms=0.078) -> cutoff=45654 Hz | wall=False | ratio=0.487 | peak=88.9 | NR=False
[15:09:34] slot@374s(rms=0.18) -> cutoff=46358 Hz | wall=False | ratio=0.233 | peak=97.8 | NR=False
[15:09:36] end@420s -> cutoff=45889 Hz | wall=False | ratio=0.412 | peak=80.7 | NR=False
[15:09:36] [OK] Spectrogram Hi-Res: genuine content | cutoff ~46358 Hz (> 28 kHz)
[15:09:36]
[15:09:36] [2/3] C:\Users\Vladimir\Desktop\12. Darude - Sandstorm384.flac
[15:09:37] Full analysis (>= 170 kbps) │ Engine: spectrogram (Hi-Res 384000 Hz - auCDtect bypassed)
[15:09:37] No encoder tag │ flac │ 384 kHz │ 2ch │ 24 bit │ 7652 kbps │ 00:07:23
[15:09:47] slot@89s(rms=0.009) -> cutoff=92340 Hz | wall=False | ratio=0.415 | peak=81.5 | NR=False
[15:09:49] slot@157s(rms=0.151) -> cutoff=91965 Hz | wall=False | ratio=0.586 | peak=133.3 | NR=False
[15:09:50] random@251s(rms=0.078) -> cutoff=92152 Hz | wall=False | ratio=0.502 | peak=118.1 | NR=False
[15:09:52] slot@374s(rms=0.18) -> cutoff=91965 Hz | wall=False | ratio=0.59 | peak=126.4 | NR=False
[15:09:53] end@420s -> cutoff=92340 Hz | wall=False | ratio=0.421 | peak=87.6 | NR=False
[15:09:54] [OK] Spectrogram Hi-Res: genuine content | cutoff ~92340 Hz (> 28 kHz)
[15:09:54]
[15:09:54] [3/3] C:\Users\Vladimir\Downloads\Trance Classics HD Vinyl Rips\12. Darude - Sandstorm.flac
[15:09:54] Full analysis (>= 170 kbps) │ Engine: spectrogram (Hi-Res 192000 Hz - auCDtect bypassed)
[15:09:54] Encoder: fre:ac v1.1.6 (20230210) │ flac │ 192 kHz │ 2ch │ 24 bit │ 5639 kbps │ 00:07:23
[15:10:12] slot@89s(rms=0.009) -> cutoff=96000 Hz | wall=False | ratio=0 | peak=79.8 | NR=False
[15:10:13] slot@157s(rms=0.151) -> cutoff=96000 Hz | wall=False | ratio=0 | peak=110.7 | NR=False
[15:10:14] random@251s(rms=0.078) -> cutoff=96000 Hz | wall=False | ratio=0 | peak=98.8 | NR=False
[15:10:15] slot@374s(rms=0.18) -> cutoff=96000 Hz | wall=False | ratio=0 | peak=106 | NR=False
[15:10:17] end@420s -> cutoff=96000 Hz | wall=False | ratio=0 | peak=82 | NR=False
[15:10:18] [OK] Spectrogram Hi-Res: genuine content | cutoff ~96000 Hz (> 28 kHz)
[15:10:18]
[15:10:18]
[15:10:18] =========================================================================================================
[15:10:18] SUMMARY
[15:10:18] =========================================================================================================
[15:10:18] Files found : 3 (including hidden)
[15:10:18] Analyzed : 3
[15:10:18] Skipped : 0
[15:10:18] OK : 3
[15:10:18] FAKE : 0
[15:10:18] SUSPECT : 0
[15:10:18] UNKNOWN : 0
[15:10:18] Orphan dirs : N/A (file-mode scan)
[15:10:18] ===================================================================================================
The tool is working correctly and the results are exactly what you'd expect.
The original 192 kHz rip (file 3) shows cutoff=96000 Hz with ratio=0 across all slots, that means the spectrum fills precisely up to the Nyquist limit of 192 kHz with a perfectly flat wall, which is the normal fingerprint of a genuine hi-res capture. The ratio=0 here means there's no energy drop before the ceiling, which at exactly Nyquist is expected and not a red flag.
The 96 kHz downsample (file 1) shows cutoff around 46 kHz, correct, since Nyquist for 96 kHz is 48 kHz, and a low-pass filter applied during downsampling typically rolls off a bit before the hard ceiling.
The 384 kHz upsample (file 2) shows cutoff around 92 kHz, also correct. Upsampling doesn't create new ultrasonic content, so the energy stops at roughly the same point as the original (~96 kHz), which sits well below the new Nyquist of 192 kHz. The tool still passes it as OK because it sees genuine broadband content, just not artificially extended to fill the new ceiling.
So all three verdicts are correct. The tool isn't fooled by the up or downconversion.
Edit: Good experiment. The tool correctly identifies all three files, but there's an important caveat for the 384 kHz version: at that sample rate the detection is not FAKE but SUSPECT, because there's no hard frequency fingerprint to pin the source down. What the tool sees is that the content stops at ~96 kHz, which is only 50% of the available Nyquist (192 kHz), consistent with an upsample from 192 kHz, but not provably so. Support for this case has not been added in yesterday's update.
Edit 2: Tool updated 2026-06-24.
Last edited:
Julian Del Agranda
Elite Member
- Jul 3, 2020
- 2,327 Posts
- 3,189 Thanked
Good idea, discussions should continue on official forum on Audio Science Review
Audio Fake Detector PRO
The disk files for reviewers are the same ones that are sold on platforms and the same ones used for CDs. It's not a destructive marking like the one on films sent to reviewers or Oscar or Cesar voters. Interesting. You seem to be implying that if you rip CDs (as I do), you will pick up...
www.audiosciencereview.com
Last edited:
- Status