Compress Audio File Size Without Losing Quality: A Practical Guide

To compress audio file size, re-encode the recording at a lower bitrate in an efficient lossy format such as MP3, AAC or Opus, and switch to mono if the recording is mostly one voice. Bitrate drives file size, so going from an uncompressed WAV (about 1,411 kbps) to a 96 kbps MP3 makes the file roughly 15 times smaller, and for speech you will rarely hear the difference. The quickest way is to upload the file to the free WaveToWord audio compressor, pick Light, Balanced or Strong, and download a smaller MP3. If you would rather work offline, the ffmpeg, Audacity and Apple Music steps below get the same result.

The rest of this guide explains why that works and what setting to use for each job. It covers how bitrate, sample rate and channels add up to file size, how to work out the final size before you export, and the mistakes that leave you with a file that is either still too big or sounds broken.

Key takeaways

• File size = bitrate × duration. One hour at 128 kbps is about 57.6 MB; at 64 kbps it is about 28.8 MB.
• Bitrate is the main lever. Sample rate and bit depth only matter for uncompressed formats like WAV.
• Spoken word sounds fine at 64-96 kbps in MP3, and even lower in Opus. Music needs 128 kbps or more.
• Mono halves the data a voice recording needs. Use it for podcasts, lectures and voice memos, not for music.
• Always compress from the best original you have. Compressing an already compressed file again loses quality without saving much space.

Why audio files get so big

A digital recording is a long list of numbers. Each number (a sample) describes the air pressure at one instant, and there are thousands of them every second. Three settings decide how much data that list takes up.

• Sample rate: how many samples are taken each second. CD audio uses 44,100 (44.1 kHz). Video and most recording apps use 48 kHz. Studio files are often 96 kHz.
• Bit depth: how much detail each sample stores. CD audio is 16-bit. Studio recordings are often 24-bit or 32-bit float.
• Channels: mono is one channel, stereo is two. Stereo doubles the data.

Multiply those together and you get the bitrate of uncompressed audio. For CD-quality stereo that is 44,100 × 16 × 2 = 1,411,200 bits per second, or about 1,411 kbps. That is what a WAV or AIFF file stores, and it explains why a three-minute WAV song weighs around 30 MB and a one-hour WAV interview runs past 600 MB.

Lossless vs lossy compression

There are two very different ways to shrink that data.

Lossless compression (FLAC, ALAC, WavPack) packs the samples more cleverly, the way a ZIP file does, and gives you back every original bit when you play it. It typically saves 30-60% for music, depending on how complex the material is. That is great for archiving masters, but it will not turn a 600 MB recording into something you can email.

Lossy compression (MP3, AAC, Opus, Vorbis) throws away information you are unlikely to hear. These encoders use psychoacoustic models: they spend bits on the sounds your ear is most sensitive to and drop detail that is masked by louder sounds or sits at frequencies where hearing is weak. This is how a 1,411 kbps stream becomes a 128 kbps file (about 91% smaller) that most listeners cannot tell apart from the original on ordinary speakers.

When people search for how to make an audio file smaller, lossy compression is almost always the answer. Lossless formats are the right tool only when you need a perfect copy.

How to compress an audio file: the file size math

Once a file is lossy-encoded at a fixed (constant) bitrate, its size is easy to predict. The formula is:

Size in megabytes = bitrate in kbps × duration in seconds ÷ 8 ÷ 1,000

You divide by 8 because there are 8 bits in a byte, and by 1,000 to move from kilobytes to megabytes. Here is a worked example for a one-hour podcast episode (3,600 seconds):

1. At 128 kbps: 128 × 3,600 = 460,800 kilobits. Divide by 8 to get 57,600 kilobytes, or about 57.6 MB.
2. At 96 kbps: 96 × 3,600 ÷ 8 = 43,200 KB, or about 43.2 MB.
3. At 64 kbps: 64 × 3,600 ÷ 8 = 28,800 KB, or about 28.8 MB.
4. As uncompressed 16-bit, 44.1 kHz stereo WAV: 1,411.2 × 3,600 ÷ 8 = 635,040 KB, or about 635 MB.

You can also run the formula backward to hit a size limit. Say you need a 40-minute (2,400-second) lecture under 20 MB. Turn 20 MB into kilobits: 20 × 1,000 × 8 = 160,000. Divide by 2,400 seconds and you get about 66 kbps. Pick the next standard bitrate below that (64 kbps) and leave a little headroom for the file's metadata and cover art.

Expect real files to come out slightly larger than the formula says. Embedded album art, ID3 tags and container overhead add a few kilobytes to a few hundred kilobytes. Variable bitrate (VBR) files also average out above or below their target, depending on the material.

File size by format and bitrate

The table below shows how big one hour of audio is at common settings, and what each setting is good for.

FLAC sizes vary with the content, so treat that row as a rough range. Every other lossy row follows straight from the formula above.

Which codec should you use?

Codec choice changes how good a file sounds at a given bitrate, and whether the person you send it to can play it.

MP3

MP3 is the most compatible audio format there is. Every phone, car stereo, browser, editor and smart speaker plays it. Quality is very good at 128 kbps and above for music. For voice it holds up down to 64 kbps, especially in mono. Below about 48 kbps, MP3 starts to sound swirly and metallic. When you do not know what the listener will use, MP3 is the safe choice.

AAC (M4A)

AAC was designed as MP3's successor and gives slightly better quality at the same bitrate, most noticeably at 128 kbps and below. It is the default on Apple devices and is used inside most MP4 videos. Files usually have an .m4a extension. Compatibility is excellent on modern devices, though some older hardware players and car systems handle only MP3.

Opus

Opus is an open, royalty-free codec standardized by the IETF (RFC 6716). It is the most efficient general-purpose lossy codec in wide use. Speech at 24-32 kbps in mono sounds clean, and music is very good around 96-128 kbps. Browsers, Android, WhatsApp-style voice messaging and video calling apps use it heavily. Its weakness is support in older software and some hardware players, so it is great for your own archive or web playback and less ideal as an attachment for someone who might open it in an old media player.

Ogg Vorbis

Vorbis (usually in an .ogg file) is Opus's predecessor. It is still common in games and some open-source software. For new files, Opus generally sounds better at the same size.

Constant vs variable bitrate

Constant bitrate (CBR) spends the same number of bits every second, so the size is exactly predictable. Variable bitrate (VBR) spends more bits on complex passages and fewer on silence or simple sounds, which usually gives better quality per megabyte. Use CBR when you must hit an exact size. Use VBR when you care more about quality and are happy with an approximate size. For the LAME MP3 encoder, VBR quality setting 2 averages around 190 kbps and setting 5 averages around 130 kbps, according to the ffmpeg MP3 encoding guide.

Recommended settings by use case

There is no single best setting. A voice memo and a mastered album need very different treatment. Use this as a starting point.

Podcast or interview (voice)

• Format: MP3 for distribution, since every podcast app supports it.
• Bitrate: 64 kbps mono for voice-only shows. 96-128 kbps stereo if you have music, sound design or stereo-panned hosts.
• Sample rate: 44.1 kHz or 48 kHz. Leave it as recorded.

Music

• Format: AAC or MP3 for sharing, FLAC for keeping.
• Bitrate: 128 kbps is acceptable, 192-256 kbps is safer for critical listening. Never go mono, since you would collapse the stereo image.

Audio for transcription

Speech recognition engines do not need high-fidelity audio. Many work internally at 16 kHz mono; the transcription pipeline in WaveToWord, for example, converts every upload to a 16 kHz mono WAV before the speech model reads it. That means you can compress a long recording to 64 kbps mono before uploading and lose nothing that matters to the transcript. What does matter is clean speech: avoid going so low (say, under 32 kbps MP3) that consonants smear, because "s", "f" and "th" sounds carry a lot of meaning. Once it is small enough, you can upload it to the audio to text tool. Our guide on how to transcribe audio for free walks through that next step.

Email, WhatsApp, Discord and other chat apps

Many email services cap a message at around 20-25 MB; Gmail, for example, allows up to 25 MB of attachments per message. Base64 encoding inflates attachments by about a third in transit, so aim for files well under the cap; around 15 MB is a safe ceiling for a single email attachment. At 64 kbps mono, about 30 minutes of speech fits in roughly 14.4 MB.

Discord's free upload limit is small and has changed more than once, so check Discord's current limit before you export. Use the formula to plan: a 10 MB limit holds about 20 minutes at 64 kbps, and a 20 MB limit about 40 minutes. Messaging apps such as WhatsApp accept much larger files sent as documents, but smaller files send faster on mobile data and are easier for the recipient to play. For video clips and their limits, see our guide on how to make a video file smaller.

How to compress audio file size with WaveToWord

The WaveToWord audio compressor is built for the common case: you have a recording and need it smaller for an email, a chat app or an upload form. It is free with a free account, and you can run it up to 20 times a day.

1. Open the tool. Go to the Audio compressor and sign in with your free account.
2. Upload your file. It accepts MP3, WAV, M4A, AAC, OGG, Opus, FLAC, WMA and WebM audio. Signed-in accounts can upload files up to 500 MB, and the tool page shows your current limit.
3. Choose a compression level. The tool shows an estimated output size for each level before you start:
   • Light: 128 kbps stereo MP3, near-original quality.
   • Balanced (the default): 96 kbps stereo MP3, good quality and much smaller.
   • Strong: 64 kbps mono MP3, the smallest file, ideal for speech.
4. Start and wait a few moments. Processing runs on the server, so you can keep using your browser.
5. Download the result. The page shows the original size next to the new size. The file is named after your original with "-compressed" added, and your uploads and results are deleted automatically after about an hour.

The output is always MP3, the most widely compatible choice, so the file will play anywhere. The tool keeps the first audio track of the file, so if you upload a WebM with audio, only the sound is kept. If you need a different output format, such as Opus for an archive or M4A for Apple devices, use the audio converter, which writes MP3, WAV, M4A, OGG, FLAC or Opus. And if your recording has long dead air at the start or end, trimming it with the audio cutter first is the cheapest compression there is: every minute you remove is a minute you do not have to store.

A practical tip: if Strong is still too big, the issue is duration, not settings. Split the recording into parts with the audio cutter, or use the size formula to decide how long each part can be.

Other ways to compress audio

You may prefer a desktop tool for batch work, offline privacy, or exotic settings. These are all free.

ffmpeg (Windows, macOS, Linux)

ffmpeg is the open-source engine behind most media tools, including WaveToWord's. It runs from the command line and gives you complete control. Install it from ffmpeg.org (or brew install ffmpeg on macOS), then use these commands.

Convert a WAV to a 96 kbps constant bitrate MP3:

ffmpeg -i interview.wav -c:a libmp3lame -b:a 96k interview.mp3

Make a small mono MP3 for speech:

ffmpeg -i lecture.m4a -ac 1 -c:a libmp3lame -b:a 64k lecture.mp3

Use VBR for better quality per megabyte (quality 5 averages about 130 kbps):

ffmpeg -i song.flac -c:a libmp3lame -q:a 5 song.mp3

Encode speech to Opus at 32 kbps mono, which is tiny and still clear:

ffmpeg -i memo.wav -ac 1 -c:a libopus -b:a 32k -application voip memo.opus

Encode to AAC in an M4A file for Apple devices:

ffmpeg -i podcast.wav -c:a aac -b:a 128k podcast.m4a

Batch-compress every WAV in a folder (macOS, Linux, or Windows with Git Bash):

for f in *.wav; do ffmpeg -i "$f" -ac 1 -c:a libmp3lame -b:a 64k "${f%.wav}.mp3"; done

Audacity

Audacity is a free, open-source audio editor for Windows, macOS and Linux. Open your file, then choose File > Export Audio (older versions say Export as MP3). Pick MP3 as the format, choose a bitrate mode (Constant for predictable size, Variable or Preset for quality), set the quality, and pick Mono under channels if the recording is a single voice. Audacity is also handy because you can cut silences and normalize the level before you export.

Apple Music app (Mac) and iTunes (Windows)

Apple's Music app can convert files to smaller AAC or MP3 copies. Open Music > Settings > Files > Import Settings, choose AAC Encoder or MP3 Encoder, and pick a quality preset or custom bitrate. Then select a track in your library and choose File > Convert > Create AAC Version (or MP3 Version). The new copy appears next to the original. iTunes on Windows has the same option under Edit > Preferences > General > Import Settings.

QuickTime Player (Mac)

If you only need the audio from a recording or movie, QuickTime Player's File > Export As > Audio Only saves an M4A (AAC) file. You do not get bitrate control, but for a quick, smaller voice file it works.

On your phone

Voice recorder apps usually let you choose a quality setting before recording. On iPhone, the Voice Memos app has an audio quality option in Settings > Apps > Voice Memos: choose Compressed rather than Lossless to save space on future recordings. Recording smaller from the start beats compressing later. For existing memos, the browser-based tools above work on mobile too, and our guide to turning a voice memo into text covers what to do next.

Common mistakes that hurt quality

Compressing a file that is already compressed

Every lossy encode throws information away. If you take a 128 kbps MP3 and re-encode it to 96 kbps, the encoder works on audio that already has artifacts, and it adds its own on top. This is called generation loss. The file does get smaller, but quality drops faster than it would have from a clean original. Whenever possible, compress from the original WAV, FLAC or high-bitrate recording.

Raising the bitrate to "improve" quality

Converting a 64 kbps MP3 to 320 kbps makes the file five times bigger and sounds exactly the same, or very slightly worse. Lost information cannot be restored. A higher bitrate only helps if the source actually contains more detail.

Using mono for music

Downmixing a stereo song to mono halves the data, but it also throws away the stereo image, and some effects can cancel out when the left and right channels are summed. Mono is perfect for speech and a poor trade for music. If you need a smaller music file, lower the bitrate a little or switch codec to AAC or Opus instead.

Lowering the sample rate too far

For lossy formats, bitrate is the setting that controls size. Lowering the sample rate of a lossy file does not shrink it much by itself. Going very low (like 8 kHz) produces telephone-quality sound because it cuts off everything above 4 kHz. If you want speech-optimized files, keep 44.1 or 48 kHz and lower the bitrate, or use Opus.

Choosing a format the listener cannot play

A tiny Opus file is useless if your client opens it in software that does not support it. When in doubt, send MP3.

Ignoring silence and length

A recording with ten minutes of setup chatter at the start is ten minutes of wasted megabytes. Trim first, compress second.

Troubleshooting

The compressed file is barely smaller than the original. Your source was probably already a low-bitrate MP3 or AAC. Check its bitrate in your file properties (on Windows, right-click > Properties > Details; on Mac, select the file and use Get Info, or open it in QuickTime and choose Window > Show Movie Inspector). If it is already 64-96 kbps, the only real way to save more space is a lower setting, mono, Opus, or shortening the recording.

The file got bigger. You converted a compressed file to a higher bitrate or to a lossless format. Use a bitrate at or below the source's.

The audio sounds watery, metallic or "underwater." The bitrate is too low for the material, especially music with cymbals or reverb. Step up one level (for example, 64 to 96 kbps), or switch to AAC or Opus, which handle low bitrates more gracefully than MP3.

One side of the stereo recording was silent, and now the whole file sounds quieter. When one channel is empty, a mono downmix averages it with the live channel and can lower the level. Normalize or amplify in an editor such as Audacity before exporting.

My email or chat app still rejects the file. Remember that attachment limits apply to the encoded message, and many apps also count every file in the message together. Aim for about 70% of the stated limit, or share a cloud link instead.

The upload to an online tool fails. Check the size limit shown on the tool page, try a stable connection, and make sure the file plays on your computer. A corrupted or partially downloaded file cannot be decoded by any compressor.

Lossy or lossless: a quick decision guide

• Keeping the master of your podcast, music or field recording? Save it as WAV or FLAC and never delete it.
• Sharing with listeners, colleagues or clients? MP3 at 96-128 kbps (or 64 kbps mono for speech).
• Archiving hours of speech where space matters and you control the playback? Opus at 24-32 kbps mono.
• Sending something to transcribe? MP3 at 64 kbps mono is more than enough.

Think of the lossless master as your negative and the lossy copies as prints. You can always make a new print at a different size, as long as you keep the negative.

If your audio is trapped inside a video, you will need to pull out the soundtrack first. Our guide on how to convert video to audio explains how, and the WAV vs MP3 comparison goes deeper into when each format makes sense.

Compress your audio file now

You do not need to install anything or memorize ffmpeg flags to get a smaller file. Open the free WaveToWord audio compressor, sign in with your free account, upload your recording, and choose Light, Balanced or Strong. You will see the estimated size before you start and the real size when it finishes, and you get an MP3 that plays on every device. When the file is small enough, you can send it, upload it, or turn it into a transcript with audio to text.