A high-bit lossy file over, say, 320 kbps, will not show any signs of being lossy if it is converted to a lossless container like FLAC. When you get up into the 400s, you have to start wondering how anal you are going to be about it.
tuttle touched on this. No upsampling a file will fill the gaps with dithered data. So it also file size scales to the upsampling. JPEGs will do the same thing except with more subtlety with enlarging a photo, since the largest portion of a JPEG is palette. With dithering a JPEG, the palette increases because dithering by filling in a gradient between the pixels. This dithering occurs on the fly when playing lossy files. Upscaling hard codes the dither to new lossless file.
Most of what you wrote here is a creative work of fiction. Not even layperson explanation, just wrong. JPEG does not have a palette map. It is a 24 bit format, 8 bits per color. JPEG does not "do the same thing", or really do anything, when you resize. Image manipulation software does the resizing, outside of the compression codec or container. There is no dithering in JPEG, because again, there is not a bigger bit depth being converted into a smaller bit depth. With psychoacoustic lossy codecs, there is also no "dithering" or "upscaling" or anything else you mention... Lossy codecs generally transform audio from the time domain into the frequency domain with FFT, at the same general data rate and quality as the source bitrate. Then they do analysis, using models of human hearing, to determine when certain frequencies are masked or inaudible. The bins of inaudible frequencies are then not encoded or stored in the lossy container if they are not contributing to the perceptible experience. When the lossy file is played back (decoded), it is simply turned back into the time domain with the non-encoded information not present. The output is still the same sampling rate and bit depth (although actually in MP3, the floating point numbers that store the audio can be decoded to 24 bit PCM for more accuracy). On medium-fidelity perceptually-encoded files, using frequency analysis with the correct windowing and small number of bins (like 1024 in this example) one of the tell-tale signs is the "grittiness" or "holes" that one can see in the spectral view: Or with more time resolution: Another sign is, if you use an extremely small number of bins for high temporal resolution, your view doesn't actually give more resolution, because the audio has already gone through a differently-sized FFT transformation. *highest bitrate reliably decoded by all general-purpose devices and decoders. It actually supports larger frames than 320kbps, and the data can be spread between frames with the frame buffer.
That depends on what album you are talking about. Awhile back I wanted a particular album and the used, legit CD copies sell from $75-$100 because the album is OOP and in demand with certain types of collectors. I purchased the FLAC from the Tidal Store for $18. Of course I checked it with Spek as soon as I downloaded it, and luckily it was legit. Some of use also follow independent artists on Bandcamp. I made a thread awhile back detailing experiences where certain labels or artists that do not know better will upload lossy files to their store. Spek is also very useful in that case.
This application runs on a Mac. It even tells you if the Flac file was upsampled. As you can see, it detects the bitrate on two tracks that don't match.
So, what's the name of this software? "Fakin' The Funk"? Great name! Fakin' The Funk? - Detect the true quality of your audio files in one batch. Yes, it also runs on Windows and Linux. Downloading it now!
See, I bought an Aphex Twin album with bonus tracks as a digital download in the form of FLAC files and because I have no life I checked the bonus tracks in a spectrogram because they were previously released as mp3 files and the frequency cut off waaaay below 20khz. Like, there was hardly any content above 5khz. Out of fear I was cheated I checked all of the other tracks and same result. I have an original FLAC rip of the same album on CD and those songs too, were really really below 20khz. The music itself is mostly ambient electronic sounds, like this: I'm completely clueless about the nature of recording music. Is this just the nature of electronic music? Aphex Twin's music is pretty lo-fi by nature and he more than likely recorded this album to DAT tapes. But can lossless music by nature be this "lo-fi"? I can't tell
Funny thing is whenever I buy anything from Bandcamp (underground black metal, admittedly) it says I'll have my choice of FLAC or mp3 but after paying up only the shoddy 128-320 VBR mp3 appears as the sole option. Why is this ?
I always use mid/side as a general rule of thumb for those questionable sounding flac files. Usually the side channel will suffer more loss in the higher frequencies and they will tend to look a bit different on a spectrogram. I think it has to do with the encoding scheme of each different lossy audio compression algorithm. This has sorta been discussed here already in another thread. As stated before use your ears, you can usually hear those unwanted odd sounding artifacts in the song.
I've ripped my legit CDs to FLAC and audiochecker never tells me that it's close to 100% sourced from a CD. Not sure I'd trust the application.
Not necessarily. As an experiment, I ripped some audio from YouTube videos and the output was set for .flac. For example, Luther Ingram's "(If Loving You Is Wrong) I Don't Want to Be Right" was sourced from MPEG, but it put out a file with a bitrate of 798kbps and a size of 20.59mb.
Another way is to analyze the file with Audiochecker from Softpedia, which will return a result identifying the probable source and a percentage of probability, i.e. "70% CDDA" or "98% MPEG."
The MP3 format in itself does not cut off high frequencies. It supports sampling rates of 44.1 or 48 kHz amongst others, so the full audio range up to 20 kHz. It does compromise quality, but not in this crude way, and if done right it's barely or not at all audible. So if the audio of your recordings has few highs, there is something else going on. Either this was an artistic choice, or a mastering decision. Do you have access to an original CD release for comparison? No
This is not what was being asked here. The OP wants to make sure that his FLAC files came from a pure PCM source (like a CD or a WAV file ripped from it), and not from MP3s that were blown up again. Your code merely checks if the files technically adhere to the FLAC format, which they will in either case. The answer is to compute the difference waveform between the FLAC in question and an original CD known to be from the same mastering. If the difference is zero (total silence), then they are from the same source as the CD. If the difference is some weird residue, that somehow contains artefacts of the original music, then the FLAC might have been upconverted from an MP3. But there are other reasons why this might happen (different mastering, or incorrect alignment during the computation), so it's not necessarily a bad sign.
Some lo-fi and lo-fi ambient music can be like that due to the equipment used during recording and mixing. Some of the older synths weren't exactly hi-fi (like the old Casio synths and such). Some of the older digital effects boxes back then were also not hi-fi. If equipment like that was used in making the recording or mixing the recording then you'll end up with higher frequencies chopped off and other artifices that might make you think it is from a lower bitrate MP3 source rather than the actual studio master. I've got the 1994 CD release of Aphex Twin "Selected Ambient Works Volume II" on CD. I don't have the 2017 re-release with bonus tracks to compare. I ran CD1 and CD2 through the MPEG analyser in Trader's Little Helper. Here's the results: Code: auCDtect: CD records authenticity detector, version 0.8.2 Copyright (c) 2004 Oleg Berngardt. All rights reserved. Copyright (c) 2004 Alexander Djourik. All rights reserved. Detect mode (0..40 with 0 = most accurate): 8 (default) 01 - [Cliffs].wav: track looks like MPEG with probability 89%. 02 - [Radiator].wav: source of track cannot be qualified. 03 - [Rhubarb].wav: track looks like CDDA with probability 76%. 04 - [Grass].wav: track looks like CDDA with probability 94%. 05 - [Mold].wav: track looks like CDDA with probability 94%. 06 - [Curtains].wav: track looks like CDDA with probability 62%. 07 - [Blur].wav: track looks like CDDA with probability 92%. 08 - [Weathered Stone].wav: track looks like CDDA with probability 94%. 09 - [Tree].wav: track looks like CDDA with probability 86%. 10 - [Domino].wav: track looks like CDDA with probability 54%. 11 - [White Blur 1].wav: track looks like MPEG with probability 99%. Code: auCDtect: CD records authenticity detector, version 0.8.2 Copyright (c) 2004 Oleg Berngardt. All rights reserved. Copyright (c) 2004 Alexander Djourik. All rights reserved. Detect mode (0..40 with 0 = most accurate): 8 (default) 01 - Blue Calx.wav: track looks like CDDA with probability 100%. 02 - [Parallel Stripes].wav: source of track cannot be qualified. 03 - [Shiny Metal Rods].wav: track looks like CDDA with probability 100%. 04 - [Grey Stripe].wav: source of track cannot be qualified. 05 - [Z Twig].wav: source of track cannot be qualified. 06 - [Windowsill].wav: track looks like CDDA with probability 100%. 07 - [Hexagon].wav: track looks like CDDA with probability 87%. 08 - [Lichen].wav: source of track cannot be qualified. 09 - [Spots].wav: track looks like CDDA with probability 100%. 10 - [Tassels].wav: source of track cannot be qualified. 11 - [White Blur 2].wav: source of track cannot be qualified. 12 - [Matchsticks].wav: source of track cannot be qualified. Disc 1 looks like a lot of the tracks are from lossy source or poor quality lossless source. Disc 2 is better. But still makes you wonder if there is some lossy source in there. That's just the nature of the recording and how it was recorded and mixed. I very very much doubt that they submitted lossy files to the mastering engineer. Back in 1994 lossy quality would have been much worse than is evident in this analysis.
I know, but his title question was vague and confused. If you use the "Exact Audio Copy V1.3" for the original CD ripping, there is a log file with track by track confirmations (eg. Accurately ripped (confidence 200) [FA208BB0] (AR v2) Copy OK...). You can see there if your wav copy is exact like the CD original wav. Or not. The EAC app. is the best CD ripping tool.
If an artist recorded some parts of the album at home using some early digital home studio style audio gear that may have recorded to a lossy format (back then storage was very expensive so some equipment did record to lossy in order to save space or limited processing ability). So the artist records part of the album to a lossy format. Sends that to mixing where it gets mixed and has some reverb effects added. Then that gets sent to mastering where it gets the final EQ and sent to CD replication. Parts of what is on that CD come from poor quality early digital or even lossy processing. You can rip that CD and EAC will say it is verified and a perfect rip. But the fact is that parts of what's on that CD have come from a lossy source. The editing and changes made during mixing and mastering will make that lossy-ness less obvious. But it's still partly a lossy source. And some analysis tools will pick up on that and flag it as a potentially lossy source.
I use spek and audiochecker. I have purchased numerous FLAC downloads over the years and all have proved to be legit. To test audiochecker, I upsampled about 500 tracks. Audiochecker reported that they were all lossy sourced. Not one of them was reported as lossless. I trust audiochecker.
It's about files he bought online, not ones he created himself. In that case he wouldn't need to ask.