Encoding to a “lossy” format such as MP3 ALWAYS reduces the sound quality irreversibly.
Re-encoding an MP3 file will reduce the quality further. It’s like photocopying - each “generation” is a bit worse than the previous one until it eventually becomes a blurry mess.
Higher bit rates will cause less loss of sound quality than low bit rates, but all will reduce the quality to some extent.
Converting to a “lossless” format such as WAV will preserve the same quality as the original file.
Thank you for this answer. I also looked at the FAQ but didn’t find answers to some of my specific questions.
I understand about mp3 being lossy. The radio station equipment will not play WAV. For now, let’s assume I must use mp3, and I want to know the best way to use it, so I need to understand certain principles involved.
If I download a file that is 128, and open it in Audacity, what is the best rate to export the edited file? Is it better to export it at 128, or would exporting at 320 distort it more, being a different rate?
If I instead use Audacity to record the same item playing at 320, and after editing I export it at 320, would that be better than the downloaded 128, or does going the record route lower the quality more?
The station’s internet stream is only 128, in case that matters.
With the LAME MP3 encoder, you can use variable bitrate and then you choose the quality (i.e. V0 = best quality) and LAME determines the required bitrate for the particular material… Some material is easy to compress, and other material is harder.
It gets a little tricky… You cannot say that 320kbps is “better” than 128 or 192kbps if they all sound identical, or if they all sound identical to the uncompressed original (which is possible, depending on the program material and the listener’s ability to hear compression artifacts).
With lossless compression (FLAC or ALAC, etc.) quality is always identical to the original. The bitrate depends on the original-uncompressed bitrate, how “smart” the compression algorithm is, and how hard it “works”. i.e. FLAC has several settings that determine how much processing it does and how much time it takes to compress.
With uncompressed formats (WAV, etc.) the bitrate (bitrate kilobits per second) is simply determined by the number of samples per second (kHz), the number of bits per sample, and the number of channels. i.e. An audio CD is 44.1kHz x 16 bits x 2 channels = 1411kbps.
***** As long as you know that there are 8 bits in a byte, you can calculate file size (of any audio or video file) from the bitrate & playing time. (With a little uncertainty due to tags/metadata, the file header, and other “overhead”.)
You can minimize the additional damage by saving at the highest bitrate you can. If you just make another 128 clip, it will have sound damage similar to 64. If you export 384 or higher, many people won’t be able to hear the damage increase, but, of course, the size will go way up.
If you’re doing very simple editing, you might be able to use one of the native MP3 editors and not use Audacity at all. There will be no increase in damage because those editors don’t take the original MP3 apart and need to make a new one.
The radio station equipment will not play WAV.
That’s rough to believe. Maybe they won’t accept WAV files from outside contributors.
I guarantee if you have the only WAV file of shots fired at the local high school, somebody would figure out a way to play it.
So if you give them a 320 kbps MP3 then they would have to convert it to 128 kbps.
If they can convert a 320 kbps MP3 to their 128 kbps stream, then surely they must be able to convert WAV files?
Check with the radio station and ask them precisely what format you need to give them. The options are: CBR or VBR, mono or stereo or joint stereo, and then the bit rate (kbps).
I appreciate all the informative replies, and I’m starting to catch on. I’ll check out native MP3 editors. But there are a few points I’m not clear on, wanting to understand how digital audio works.
Since the station stream is only 128, does it help at all for me to save a 128 file at 320 instead of at 128?
At the station, we use CD players with flash drive sockets, so I bring my files on a flash drive. Wav files will not even register on the readout, so i can’t even find them to select. Making CDs from WAV might work, but this would be very time-consuming.
Assuming I could find a way to play WAV: If I open an mp3 file in Audacity, will there be no loss in the conversion to WAV? Or would I need to start with a WAV file to get the benefit?
What about using Audacity to record a stream running at a high rate? Is there any loss from using the record mode? So that if i record a 320 stream, am I better off than starting with a downloaded 128 file?
Would you say that technically the loss from converting an edited 128 file to 256 or 320 would be pretty small?
I also want to understand why the even ratio between 128 and 256 makes no difference in conversion. I imagined that it would be a more even sampling than with an odd ratio.
This whole conversion problem comes from you using MP3 as original work. If you can stop doing that, many of these quality problems will vanish. MP3 is a final step — delivery to the user for his iPod, not a step in the middle of production. Never do production in MP3.
Each time you make an MP3, some sound damage is created. That’s how it works. The magic of MP3 is its ability to cleverly hide the damage so most people can’t hear it. When you edit MP3, the new MP3 has to hide the original damage plus the new damage. The effect of this is having an MP3 in far worse a quality than you started with.
“I edited my MP3 show and exported it as the same file size, but the sound is honky and bubbly now.”
Yes. That’s correct. If the station’s stream is 128 no question, then the best you can do is shoot the work as WAV, edit as WAV and leave only one MP3 conversion at the station. 128 is the Audacity default quality and that works out very well. If you start out with 128 MP3 and cut it, then the very best you can possibly do is a final show of 64, barely enough quality for a stereo show. That’s if you use WAV (uncompressed) in the middle.
If you/they insist on using MP3 in the middle, then use the highest possible MP3 rate as an Audacity export.
With all those MP3 steps in the middle, I’d be surprised if the shows had any better quality than a cellphone.
I listen to a podcast talk show each week and I know they start with the original radio broadcast masters. It sounds perfect, but on some of the shows, they have an intro and commercial that was not part of the broadcast and they did it from multiple MP3s. It sounded nine-year-old kid awful particularly as it came right before the crystal clear show. I see they don’t do that any more.
I think I basically understand. I may still need to use mp3 if other formats won’t read on the player. And I’m largely stuck with needing to start with mp3 for much of what I use. So i need to do the best I can under the circumstances.
How does Audacity’s record feature rate in all this, if the mp3 I record is fairly high quality?
Is there any way to analyze the quality of an audio file?
And just so I’m clear: Even if I save an edited 128 file at 320, will it still end up no better than 64?
It might be a little better than a 64 kbps encoding, but the point we are trying to make is that it will be worse than the 128 kbps version that you started with. MP3 encoding always reduces the sound quality. Encoding to 320 kbps will not reduce the quality much, but it will reduce it. The lower the bit rate the more sound quality is lost and the lost sound quality is permanent - there is no way to fix it. “lossy” formats are called “lossy” because some of the information is lost (“gone and lost forever”).
Not sure what you mean. Audacity records uncompressed (lossless) data. If you want to get an MP3 file into Audacity, don’t record it, “Import” it (File menu > Import > Audio).
I’m asking, what if turn on the record in Audacity to record a streaming audio. How does recording it affect the quality of the file? If the stream is at 320, what will the recording be? I’m considering whether it is better to record the stream than to download a file if the download of the same thing is only 128, for example.
The one that plays the flash drive I use to play the audio at the radio station. The player does not register a WAV file, there is no readout. It will register an mp3 file.
But even if I were able to play WAV, the original of what I want to play is usually a 128 mp3 file, so I would need to convert it to WAV, and then the station converts to a 128 mp3 stream, which you’ve said wouldn’t be good.
At the point you record, Audacity records losslessy in PCM. The recording Audacity makes is as good or bad as the streamed MP3 is, after the stream has been converted from digital to analogue (so that you can hear it playing in an analogue sound card).
That digital to analogue conversion is technically slightly lossy, but that loss is rarely noticeable - unless for example a crackle occurs during playback, in which case the crackle will be recorded.
If the download actually is of lower audible quality than the stream, you should still record the stream, because the losses from recording should be minimal.
If the stream and download are of equal audible quality, you should download the file.
I refer to “audible quality” rather than bit rate, because a 320 kbps MP3 can sound worse than a 128 kbps MP3 if the original audio for the 320 kbps version was worse, or if the 320 kbps version was originally 128 kbps and had been re-encoded at 320 kbps. That re-encoding would make the MP3 slightly worse than when it had been 128 kbps, but not much worse.
— The 128 quality of the station’s stream is fixed. I don’t see that changing any time soon because when presented with a perfect, completely undamaged show, it produces an almost perfect web stream with the smallest server cost.
— If the delivery to the station is MP3, then it’s up to you to create the most perfect, high quality MP3 you can. And then call the station and find out about delivery in WAV. Like I said, if you have the only WAV of a major news event, I guarantee they’d figure out a way to play it.
— You took it in the shorts when you decided to repackage MP3 downloads (we assume free) instead of creating original content. Producers have good quality control over original content (as many multiple Audacity Forum posters with microphones can attest), but you are a complete slave to both the content provider and the delivery company. So the show will never be any better than the worst combination of both of those.
The info on recording is promising, though it will be more time-consuming. I try to judge the quality of the original audio I download or record. If I download an mp3 and open it in Audacity, I can see whether it was clipped, and if so I can look for one that isn’t. I also usually can tell by the graphic if it was clipped but then volume reduced. If I record, it is the same thing, i keep the volume at a level not to clip, but if I see it is flattened anyway, i look for a better copy.
I’d rather not identify the station because I don’t want to call attention to myself here personally. But it is a small non-commercial community station. I am doing music programming, using internet sources for music we don’t have in the library, which is most of it for what I do. DJs do this commonly, and the station pays the general copyright fees and publishes the playlists.
There are already sound quality issues with the broadcast and stream, I don’t think we are very good in that department. If I download a program from the archives and look at it in Audacity, it is usually clipped, which I think is because DJs don’t control their volume properly. I try to keep mine within a good range, but it is touchy.
I try to keep mine within a good range, but it is touchy.
I can explain pieces of this. The “board” or mixing desk is generally split in two. One split goes to the broadcast transmitter and through the legally required volume compressors and limiters. The other split goes to the streaming servers which everybody knows don’t really matter. Since there’s no electronics in server pathway, whatever the board operator feels like producing is what goes out. Overload, Dipping, Pumping, Clipping, etc. Out the door.
I begged one station to please get the data stream from the transmitter side of the compressors rather than the raw mixing desk so at least we didn’t get the wild volume swings and clipping the board op was producing. It worked. Their stream was spot-on day in and day out and sounded remarkably like the radio broadcast.
There are tricks to this.
Send the name of the station (assuming it’s in the US) to me as a Private Message. I’m curious about the place, and yes I do have a General Radio FCC license and know how to drive a radio transmitter.