Let's say I have a music project to export.
Is there any quality difference between:
Is there any quality difference? I understood yes, cause from MP3 to WAV it takes down a lil' of the quality due to the MP3 origin file from the conversion.
So here I am asking...
Yup that's right,
WAV is raw audio, just a long string of values, the size of each value and the amount of values you get per second being the bit depth and sample rate respectively.
Mp3 is a compressed format, one of a number of algorithms has analysed the original wav and tried to save space where possible. A simple example might be rather than holding ≈220000 24bit samples to represent 5 seconds of silence at the beginning of a single channel of an audio file (which uses 176Kb of data, note kilobytes not kilobits) an Mp3 algorithm might replace it with a statement 'value is zero here for 220000 samples', which uses vastly less data (It doesn't actually do this, it has a flag for 'inaudible' and treats silence as such but just to illustrate the idea).
Of course the techniques used to minimise file size get much much more sophisticated when audio is present in the file, and can reduce the Mp3 file size massively compared to a WAV. (something like 50Mb compared to 6Mb or so for a 3 min audio file). It gets a bit more complex when trying to generalise much further than this but it's along those sort of lines.
Depending on what bitrate your Mp3 file is you are throwing away a lot of data, and you cannot get it back. Eg if you tell your Mp3 encoder to be really aggressive and make it a 96Kbps file you'll save lots of space, but to do so the algorithm will make many more generalisations when trying to 'describe' the file for the Mp3 format. When trying to convert back to WAV you only have so much to work with, you'll essentially just end up with a raw audio, high res version of the much lower resolution Mp3 file, not the original audio. Think a high res digital photo of a low res pixelated image.
So it's better to just export anything important as wav, disk space is cheap, and doing so means a master version of a track etc. is in the correct format to justify calling it some kind of master for storage, archiving, mastering, releasing etc.
However Mp3 is a very good format (as are the newer, sometimes even better compression types such as AAC), I find it next to impossible to tell a 320Kbps Mp3 from a wav file of the same track on a good system, though some claim to. Depending on what you need to do you can still get good results. I've sampled from Mp3 for electronic music plenty of times and it's fine for that kind of use, and arguably higher fidelity than sampling from old records or cassettes which was common for decades (though the latter do arguably add a bit of noise and distortion which can be desirable for 'the sound' of those genres, the smearing and ringing of lower quality Mp3 is not considered desirable in the same way).
But unless theres a reason it's best just to stick to moving audio around as WAV or similar until you've finished recording/creating and leave Mp3 for the lightweight version of the track suited for distribution on the net or personal music collections etc, which is what the format was designed for.
Other answers give much more detail, but I think it's worth spelling out the main point:
Every time you convert to a lossy format such as MP3, you lose information i.e. quality. It's gone; you can't get it back again later.
This applies to all lossy formats/codecs: images (e.g. JPEG), audio (e.g. MP3, AAC, OGG), video (MPEG, MP4, WEBM, MKV, and most others)…
And it applies even if the data came from that same lossy format — e.g. loading an MP3 and resaving it to MP3 will lose further quality each time.*
In this case, exporting directly to WAV would give you a perfect copy of the project. Whereas exporting to MP3 would lose quality, and converting that to WAV would merely give you a perfect copy of that lower-quality version.
(* The only exception is if you're using a lossless editor — a program that can manipulate the encoded data directly, without re-encoding it. But few programs can do this; and although they can avoid losing any further quality, they can't restore the quality already lost.)
Yes, with the second option, you basically have an MP3-quality WAV file.
As Owen said, in sound design and audio production, you should only use a "lossy" format (like MP3, AAC or even a "lossy WAV" - see "side note" below) if it is required or beneficial for release/delivery. Other than that final conversion, you should have kept it in its original format. Make sure to handle conversions with dithering and antialiasing when lowering the bit depth or sample rate, respectively. Again, this should only be done last, as you can incur something called "generation loss", which is cumulative (Converting a 320 kbps MP3 to 32-bit WAV will actually (technically) LOWER the "quality", believe it or not!). Some music or sounds can be more forgiving than others of conversion and file compression artefacts, too, which is why there is no hard and fast rule. You must listen at different intensities and decide, mostly. But as a general rule,
Convert and mask a maximum of once - if necassary, and only as the last step in the chain and to a high a quality format that is possible for release.
Regarding storage space, there are "lossless" compressed formats for PCM/LPCM raw/WAVE files that just compress by searching for similar patterns and saving offsets, etc, and can be easily decoded without loss. You won't save as much space as you would with a lossy format, though (probably 50%), but you keep all your information on decoding. The only real compromise is encoding/decoding time vs storage space saved - but, as Owen said, disk space is cheap (I still use a 6 TB HDD RAID array that's been going trouble-free for years as my library storage) So what most people do, is keep the stems/mixdowns/masters as original format WAVs, and convert if and when needed.
As a side-note, the most common WAVE formats used are LPCM/PCM formats, (e.g. the humble 44.1 kHz, 16-bit encoding), which are defined in the mmreg.h header as
WAVE_FORMAT_PCM/WAVE_FORMAT_IEEE_FLOAT (0x0001/0x0003)
But the spec also allows for a lot more proprietary and non-proprietary formats, including lossy compressed formats like MP3, defined in the header file as
WAVE_FORMAT_MPEGLAYER3 (0x0055) ISO/MPEG Layer3 Format Tag.
So an MP3-encoded stream can actually be stored in the WAVE format.
So remember that ".wav" doesn't necessarily mean "PCM".
MP3 is a lossy format. MP3 encoder discards some audio information if it believes the difference would be inaudible. While high bitrate MP3 might be fine for listening, any further editing the file (e.g. mastering, sampling, mixing, remixing) might bring the inaudible differences to an audible level.
Converting MP3 to WAV doesn't recover the discarded information, therefore exporting to WAV directly allows to avoid the loss of quality.
It should be emphasized, that exporting to WAV might be lossy as well. For example, if your project uses 24 bit depth, and you export to 16 bit WAV, the last 8 bits of dynamics are lost. Moreover, dithering needs to be applied to prevent quantization noise. Again, this is particularly important if the file is meant to be edited further.
