@Polyphemus I certainly did not take it as criticism and I appreciate your observations and comments. I am trying to understand as well and learning at the same time. Whether this leads to solutions of a possible puzzle remains to be seen though.
1 Like
Awww, I did not expect that to happen 
Have a look at this. This is the same blank section, before and after amplification. You can see how the amplified part rolls off towards the centre, then ramps back up again.
Before:
And after:
Hmmm, interesting and I would like to try replicate it. What source are you using and what Amplification amount (dB) did you apply?
The source is the glitched Emily MP4 from the Game Detectives webstite (Waking Titan.MP4). I stripped out the audio to a WAV file myself using FFMPEG.
I’m inspecting it with Audacity 2.13
(edit) I think the amplification was around 30 - 36dB
For comparison, I am using a source download straight from Twitch, from which the audio was stripped using VLC in .flac format.
Here is what I get at maximum Amplification 31.109dB, no other filters/effects applied.
So yes, I do see your point. Something is however causing quite a difference and I am actually curious what that could be. I can not find the video you used as source, can you provide me a link to the page?
I can not quite get the same result. Downloaded the Twitch clip and tried several methods:
- VLC > .flac > max Amplification, basically same result as image above.
- VLC > .wav > max Amplification, basically same result as image above.
- FFMPEG > .wav > Amplification 30.856 (max). See result below
I am really not sure why I am unable to get some amplification to show in the center or close to it, even when using the same source and method. So now what? I am lost …
Firstly, it looks as if you’re including small parts of the existing audio in your sample. If you do that, amplifying it will cause the signal to clip, and it won’t amplify any more.
You also need to tell Audacity to allow clipping (it’s a check box in the “amplify” effect). Then you need to play around with different levels of amplification. It doesn’t matter if you do too much - you’re not going to play the thing, you’re only trying to look at it.
1 Like
Ok, I did indeed include edges of the audio parts and not checked the box to allow clipping.
So I tried again on the FFMPEG created wav on several sections, but same thing happens.
Before
After with 35dB ampl.
I thought science was supposed to be fun … I really do not understand this.
Edit: So first I had my reality as I observed it, then you showed me your reality as you observed it. Now we have two realities that differ and no clue which one is correct, is there a Turing test for situations like this?
1 Like
Neither do I. Here’s another one I’ve just done in the last 5 minutes
Before:
After (amplification 35.2 dB:
Is one of you using an audio format with lower bit depth perhaps? Or floating point samples versus fixed?
I am no audio expert but it seems clear that a 16 bit sample size will preserve low level signals better than an 8 bit sample size, for example.
Wow, that one is even more confusing, not even close to roll-off, more like it was supposed to be there all along. I wish I could do the same, or at least understand why we have such different results using the same method. Any ideas are welcome … unless it involves a ‘red button’.
I will stop to focus on the differences we have and just acknowledge them, while getting back on track with the initial analysis and see if it leads anywhere. I do thank you for sharing, even though I am left with new questions.
Using FFMPEG to get the audio from mp4 in .wav format preserves the quality. So a command like this should suffice: ffmpeg -i video.mp4 output_audio.wav
We both used the same video source and ffmpeg to get the audio. We both use the same version of Audacity as well 2.1.3
1 Like
There is still a roll-off there, but this is one of the shorter gaps, so the effect is less pronounced. It looks to me as though the volume has been reduced using a bathtub curve - very steep attack, short roll-off, flat bottom, short roll-up, then very steep recovery.
I will keep it in the back of my head, thank you, I do appreciate it. Do you think it could have been easily created with some code to ‘glitch’ the audio, similar to the glitching of jpeg we have seen? Debunking this as a puzzle might be easier when finding a ‘glitch’ source used. So far I am doubtful there is a puzzle, but not convinced there isn’t. Having something or any lead to proof this being a puzzle or not would be great.
Me either, I would have at least expected someone to say “what you talking about you crazy fool” lol
1 Like
did anyone try and split they audio file to left and right because they are both different wave patterns then try anything I just don’t want to waste my time if its been done
Guessing here - you could produce this sort of effect by modulating the existing audio stream with another signal. But there don’t appear to be any audio frequencies in the modulation. Just more-or-less square waves. Modulating with an inverted RF Morse signal would probably produce something like this - the RF frequency would be too high to show up on audio - except the pattern’s wrong for Morse.
You could also produce it by running the signal through a synthesiser, with the sequencer programmed to insert a specific pattern of volume reductions.
Or the video could have picked up interference from someone’s spin dryer.
(That’s a joke, really. the spacings are too regular.)