It depends what you mean by “reasonable size”, and how long the audio is.
For short files, 16-bit 44100 Hz WAV will give CD quality, which is better than any MP3 or OGG.
If that’s going to make the file size too big, then 256 kbps MP3 will give “transparent” encoding (the MP3 sounds identical to the original).
If you are streaming the audio, then best to use CBR (constant bit rate) as some players have trouble streaming VBR (variable bit rate). If the audio is for download, then “Preset: Extreme” will give the same sound quality as 256 kbps CBR, but usually in a smaller file size.
If the audio is for download, then “Preset: Extreme” will give the same sound quality as 256 kbps CBR, but usually in a smaller file size.
Most of the compressed formats will produce unperceptible or almost unperceptible internet postings. That’s what MP3 was designed to do, It’s a perception-based design. Even in smaller files, you only hear damage if you compare it directly to the original. It’s only when you get to the Perception Limit, 64 Quality Stereo, that people start to notice something odd about the sound. Audacity Export quality default for finished work used to be 128.
The real consideration for posted material is what’s going to happen to it next. MP3 doesn’t edit well. If somebody decides to convert your work to another compressed format, or re-edit the work and make a new MP3, the quality is going to take a dive. If that’s OK with you, the producer, then yes, post a master at “Insane Quality” so it’s “Good” when the client gets done.
There was a poster a while ago who produced a broadcast radio show from downloaded music (in the woods of Maine). The station insisted on submission in MP3. It wasn’t optimal but it worked OK until the station tried to make a podcast out of his show. They couldn’t do it. The third MP3 conversion turned the show to garbage.
The only thing that sample rate affects, is the available frequency bandwidth.
44100 was chosen as the standard for audio CD because it was “just enough” to capture the full audio frequency range. For the filter technology that was available at the time, 44100 Hz was arguably a little on the low side, which explains why the DVD format went a little higher with 48000 Hz. Since then, digital filter technology has improved immensely, and D/A converters can handle 20 kHz audio with ease even at 44100 Hz sample rate (not that many people can hear frequencies that high).
The video people run at 48,000. My voice shoots were almost always 48KHz, 16-bit, Stereo because my “clients” were the video production editors. I knew the productions were stereo, so I delivered the mono shoots in stereo and the tracks would drop directly into an edit with no conversion.
It’s good to know your audience.
Unless you’re a young girl, your hearing probably doesn’t go anywhere near 20,000, although there was a six foot plus tall board-op at American University that could hear reliably out to 19.2. It was said he could talk directly to bats, but that was never proven.
16bit does OK. It overloads at 0 and has quiet noise at 96. Most home recordists have a terrible time getting their shows between 3 and as quiet as 60, the AudioBook standard. 16bit easily handles that.
You should be careful not to present in a format nobody can play. 24bit is a studio standard. I don’t think my iPod can play those.
There is one more consideration. At home I have a modest internet connection. If I need to download a serious file. I have time to go make coffee. I converted several sound samples from WAV to MP3 so I could download them.