WAV is what your recorder and your editor hand you. MP3 is what a client, a host and a phone will actually take. The conversion between them is not a file extension change. It is a decision about how much detail you are willing to throw away, and it only gets made once.
Gleam Audio does the job with its Convert tool. It runs offline on macOS, Windows and Linux, and nothing is uploaded. The steps below are from version 1.1.1.
What you need
- The WAV files you want to convert, or a folder of them.
- Gleam Audio installed. Everything it needs comes inside the app, so there is nothing else to install.
- A number for the quality, which is the only setting that really matters here. The next two sections help you pick it.
If the recordings still have noise, hum or uneven volume, fix that first. Converting a noisy file to MP3 only produces a smaller noisy file. The complete guide to cleaning up spoken audio puts the order of work together.
1. Add the files and pick Convert
Drop the files on the window, or click Open Files…. Then pick Convert in the sidebar.
Everything in the list is converted in one run. That is worth knowing, because it is not how the whole app works: Cut and Split work on one file at a time, while Remove noise, Enhance and Convert take the whole list. So a session of twelve takes goes into one convert, not twelve.
2. Set the Format menu to mp3
Format is where the output type is chosen. The menu groups the choices, so mp3 sits in Common with wav, m4a, flac, ogg and opus. Studio and lossless holds AIFF, CAF, Wave64, WavPack and flac-style writers, and Other holds the rest, including AAC, OGA, WMA, AC3, MP2 and MKA.
Pick mp3 and stop there. The rest of the list is for the occasional file that has to go back to an editor in a studio format, not for delivery.
3. Set the quality, which is the real decision
Next to the format is Quality, shown in kbps, so a value like 192k. This is the one setting that decides how the result sounds, and it is the only place you get to decide it. Once an MP3 is written, no later step can put back what it left out.
For spoken word, 128 kbps is enough, and 192 kbps is comfortable. Music in stereo wants 192 kbps or more, because it has more going on across the whole frequency range than a voice does. Push a voice file well past 192 kbps and you are paying in file size for detail that nobody will hear.
Two things are worth knowing while you choose.
- Bitrate is spent on the whole file, not per channel. Stereo speech doubles the data for a difference most listeners cannot name, which is why a voice recording is a good candidate for a lower rate than a music track.
- The destination sometimes decides for you. A client who says “send me a 128 kbps mp3” has already made the choice. Check your host’s or client’s own spec page rather than guessing, because that number outranks anything you would pick by ear.
4. Choose where the MP3s go and what they are called
The bar along the bottom of the window decides the output.
- Save to puts each result next to its original by default, or in a folder you pick. The folder is remembered, along with the last five you used.
- Name sets the pattern. The default is
{name}-{tool}, and the braces button adds{name},{tool},{n}or{date}. The bar shows an example as you type, so you can see the shape of the result before you run.
Set the name before the run, not after. A folder of twelve files called by the default pattern is much easier to hand over than twelve you rename by hand.
5. Run it and watch the progress
Click Run at the top right. The button reads like Convert 3 files, so you can see the size of the job before you start it.
Each file being worked on shows how far along it is, like “Converting… 42%”, with a bar under it. If you need to stop, you have three choices:
- Stop ends the file being worked on now and skips the rest.
- Stop after this file lets the current file finish first.
- The stop button on a row ends just that file. The rest carry on.
A stopped file leaves nothing half made behind. It goes back to waiting, so you can run it again, and it does not count toward the free weekly files.
6. Check the result before it leaves your machine
When the run ends, a bar under the header says what happened, like Converted "interview.wav" to mp3., with a button to open the folder.
The Activity button, at the right of the bottom bar, lists everything you made since you opened the app. It is grouped by run, and any file still on disk can be played or shown in Finder. That is the quickest way to confirm a batch came out right without hunting through folders. If you want to run the same list again with different settings, the reset button puts finished files back to waiting.
Then check the exports the way you would check any file that is about to be published. The pre-publish checks cover loudness, the noise floor and the format, and they matter just as much on a converted file as on the master.
WAV versus MP3: what the conversion actually costs
A WAV file holds the samples as they were recorded. An MP3 holds an approximation of them, built by a model of what human hearing notices and what it can be talked out of. That is why an MP3 is often a tenth of the size of the same WAV.
The trade is one way. Going from WAV to MP3 spends information. Going from MP3 to WAV spends disk space and gets nothing back. So:
- Keep the WAV. It is the master. Convert a copy for delivery, and keep the master for the day someone wants a different edit, a different bitrate or a lossless archive copy.
- Convert once, near the end. Convert early and every later edit, level change or split happens on the lossy copy. Do the trimming, the cleanup and the leveling on the WAV, then convert.
- Pick a lossless format when the destination takes one. If the destination accepts a lossless file, opus or flac will hold the audio without the second loss. What an OPUS file is covers that choice.
When converting will not help
- Restoring quality to an MP3. Converting an MP3 to WAV produces a big file that sounds exactly like the MP3. The detail is already gone. If you only have an MP3 and it sounds thin, no format change fixes it.
- A noisy recording. Convert does not clean anything. Use Remove noise on the WAV first, then convert the result.
- Uneven volume. Convert does not level anything. A quiet episode stays quiet through the conversion. Enhance evens the loudness, and what LUFS means explains the target the destination expects.
- A clipped or echo-heavy take. Both survive the conversion, in a smaller file.
- Chapter markers inside one file. Convert writes one result per input file. If the destination wants a chapter list inside a single file, splitting the recording into parts is a different job with a different answer.
Takeaways
- Format picks the output type and Quality picks the bitrate. Quality is the decision that lasts.
- 128 kbps suits spoken word, and 192 kbps is comfortable for both speech and stereo music.
- Convert takes the whole list in one run. Cut and Split do not.
- Set Save to and Name before the run, so a batch does not need renaming afterwards.
- Watch the progress and use Stop or Stop after this file when a job has to give way.
- Keep the WAV, convert once near the end, and check the exports before they leave the machine.