Direct answer — How do you convert MP3 to text on a Mac without uploading it? Run the transcription on the Mac itself. Use whisper.cpp, a free command-line build of OpenAI’s Whisper, after converting the MP3 to a 16 kHz WAV with ffmpeg, or drag the MP3 into Weesper Transcribe, a Mac app that transcribes it on device and exports the text. The file never leaves your Mac.
A downloaded podcast episode, a voice recorder export, a saved call: MP3 is everywhere on a Mac, and turning it into text should not mean sending it to someone else’s server.
This guide shows how to convert MP3 to text on a Mac offline: what makes MP3 different, the free command-line route, the drag-and-drop route, and how to get a cleaner transcript from a low-quality file. For every other audio format, our broader guide to transcribing audio to text on a Mac offline is the place to start.
What makes an MP3 different to transcribe?
An MP3 is compressed audio that has already lost some detail. That matters when you transcribe MP3 on a Mac, because the speech model can only work with what the file still contains.
The Library of Congress describes MP3 compression as perceptual coding, “an approach based on psychoacoustic models”: the encoder discards or reduces the precision of sounds that people are less likely to hear. It notes that an MP3 at 128 kbit/s is about one eleventh the size of the uncompressed audio, and that MP3 is generally used for final, end-user delivery rather than as a master copy.
In practice, three things follow:
- Bitrate varies a lot. A podcast download, a voice recorder export and a forwarded call recording can be encoded at very different bitrates. The lower it is, the less detail the model has to work with.
- Conversion does not restore quality. Turning an MP3 into a WAV changes the container and the encoding, not the audio that was already discarded.
- Your MP3 is often a copy of a copy. Files shared through messaging apps or re-exported from an editor may have been compressed more than once.
Typical MP3 sources on a Mac
- Podcast episodes downloaded for research, quotes or show notes.
- Voice recorder exports, from a handheld recorder or a phone app set to MP3.
- Call recordings saved from a calling or conferencing tool.
- Long lectures and talks, often an hour or more of one speaker.
Why convert MP3 to text offline instead of online?
Because an online converter has to receive your file to process it. Offline MP3 transcription keeps the recording where it already is.
Most web-based “MP3 to text” converters upload the file, transcribe it on the provider’s servers and send back the result. Fine for a public podcast; for a confidential call, the audio and transcript now sit on a third party’s infrastructure, under its retention policy.
| Online MP3 converter | Offline on your Mac | |
|---|---|---|
| Where the audio goes | Uploaded to the provider’s servers | Stays on your Mac |
| Internet connection | Required for upload and processing | Not needed for transcription |
| Large or long files | Limited by upload time and the provider’s limits | Limited by your Mac and the tool |
| Data retention | Depends on the provider’s policy | Under your control |
| Setup | None | A tool or app to install once |
On-device transcription is speech-to-text that runs entirely on your own hardware, using a model stored locally. We compared both approaches in more depth in our guide to on-device vs cloud transcription.
How do you convert MP3 to text with whisper.cpp?
whisper.cpp is a free, open-source C/C++ implementation of OpenAI’s Whisper, released under the MIT licence. Its README calls Apple Silicon a “first-class citizen”, optimised through ARM NEON, Accelerate, Metal and Core ML.
It runs in Terminal and needs Git, CMake and ffmpeg. If you have never built it, our whisper.cpp setup guide walks through the installation in detail.
Step 1: build whisper.cpp and download a model
Clone the repository, download a model and build the tools, as in the project’s quick start:
git clone https://github.com/ggml-org/whisper.cpp.git
cd whisper.cpp
sh ./models/download-ggml-model.sh base.en
cmake -B build
cmake --build build -j --config ReleaseThe .en models are English-only. OpenAI notes that they “tend to perform better” for English, especially at the tiny and base sizes. For an MP3 in another language, download a multilingual model, one without the .en suffix.
Step 2: convert the MP3 to a 16 kHz WAV
The whisper.cpp README states that the whisper-cli example “currently runs only with 16-bit WAV files” and gives this ffmpeg command:
ffmpeg -i input.mp3 -ar 16000 -ac 1 -c:a pcm_s16le output.wavEach flag has a purpose. -ar 16000 resamples to 16,000 Hz, the sample rate Whisper’s own audio loader uses. -ac 1 downmixes to mono, and -c:a pcm_s16le writes 16-bit uncompressed PCM.
One nuance: the CLI’s own help output, reproduced in its README, now lists “flac, mp3, ogg, wav” as supported audio formats. Converting first remains the documented path in the main README, and it gives you a predictable input whatever build you are running.
Step 3: transcribe and choose an output
Run whisper-cli on the WAV file and pick one or more outputs: plain text (-otxt), SRT (-osrt) or VTT (-ovtt) subtitles, CSV (-ocsv) or JSON (-oj).
./build/bin/whisper-cli -m models/ggml-base.en.bin -f output.wav -otxt -osrtThe spoken language defaults to English (-l en). For a Spanish podcast, pass -l es with a multilingual model, or -l auto to let the model detect the language.
How do you convert MP3 to text with Weesper Transcribe?
Weesper Transcribe is a Mac app built for files you already have. You drag in the MP3, it transcribes on your Mac with on-device Whisper models, and you export the result. Nothing is uploaded.
It is free to download from the Mac App Store, and the free version transcribes files up to 15 minutes at full quality. It requires macOS 13 or later, on Apple Silicon or on Intel Macs with AVX2 support. It needs no microphone access, and its App Store listing says a bundled model lets it transcribe offline from first launch.
Step 1: drag the MP3 into the app
Drop the file onto the window. The app reads MP3 directly, along with WAV, M4A, MP4, MOV and any other format macOS can read, so there is no ffmpeg step.
Step 2: let it transcribe on your Mac
Transcription runs locally, with Metal acceleration on Apple Silicon. The app supports 57 transcription languages with auto-detection, so a French lecture or a German interview does not need a manual setting first.
Step 3: review the transcript
Clickable timestamps jump to the matching moment in the built-in player, handy when a low-bitrate passage is hard to make out. With Pro, you can correct names and terms directly in the text.
Step 4: export in the format you need
Copy the text, or export to one of nine formats: TXT, Markdown, SRT, VTT, DOCX, PDF, HTML, CSV and JSON. Plain text is included in the free version; subtitle and document exports are part of Pro.
Pro is a one-time purchase, not a subscription. It removes the 15-minute limit and unlocks batch processing, pro exports, editing and search. The App Store shows the current price in your region before you buy.
If you have a whole folder of MP3s, such as a season of episodes or a term of lectures, our walkthrough on how to batch transcribe audio and video files on a Mac covers the queue in detail.
Which MP3 to text route fits you on a Mac?
Choose by how much you want to do by hand. Both routes keep the MP3 on your Mac.
| whisper.cpp | Weesper Transcribe | |
|---|---|---|
| MP3 input | Convert to 16-bit WAV first (documented path) | Drag and drop the MP3 |
| Interface | Terminal | Mac app |
| Language | -l flag, English by default | Auto-detection, 57 languages |
| Export formats | TXT, SRT, VTT, CSV, JSON | TXT, Markdown, SRT, VTT, DOCX, PDF, HTML, CSV, JSON |
| Editing and search | ❌ | ✅ (Pro) |
| Mac requirement | Build from source | macOS 13+, Apple Silicon or Intel AVX2 |
| Cost | Free, MIT licence | Free up to 15 min per file; one-time Pro |
Podcasters can go further with our guide to podcast transcription and show notes on a Mac.
How do you get a more accurate transcript from an MP3?
Start from the best file you have and tell the model what it is listening to.
- Use the original export. Take the MP3 straight from the recorder or the podcast feed, not a copy forwarded through a messaging app that may have been compressed again.
- Listen to a minute first. If the MP3 sounds muffled, “watery” or clipped to you, expect more errors in the transcript and plan time for review.
- Do not re-encode to “fix” quality. Converting a low-bitrate MP3 to a higher bitrate or to WAV makes the file bigger, not clearer.
- Set the language. In whisper.cpp, the default is English: pass
-lwith the spoken language, orauto, and use a multilingual model. OpenAI notes that Whisper’s performance “varies widely depending on the language”. - Try a larger model on difficult audio. If tiny or base struggles with a noisy MP3, whisper.cpp can run the larger Whisper sizes, at the cost of speed. In Weesper Transcribe, larger models are an optional download.
- Check names and jargon before exporting. Proper nouns and specialist terms are where any speech model slips most often.
Frequently asked questions
Can I convert an MP3 to text on a Mac without uploading it?
Yes. You need a tool that runs speech recognition on the Mac itself. whisper.cpp is a free, open-source command-line option, and Weesper Transcribe is a Mac app where you drag in the MP3 and get the transcript on your Mac, with nothing uploaded.
Do I need to convert my MP3 to WAV first?
Only for the command-line route. The whisper.cpp README says whisper-cli currently runs only with 16-bit WAV files and gives an ffmpeg command for 16 kHz mono WAV, although the CLI’s help text now lists MP3. Weesper Transcribe reads MP3 directly.
Does the MP3 bitrate affect transcription accuracy?
It can. MP3 discards audio detail judged less audible, and a low-bitrate file keeps less of it. Converting to WAV afterwards does not bring that detail back, so start from the best copy you have.
Is Weesper Transcribe free for MP3 files?
The free version transcribes files up to 15 minutes, one at a time, to plain text, at full quality. Pro is a one-time purchase that removes the limit and unlocks batch processing, pro exports, editing and search.
What formats can I export an MP3 transcript to?
whisper.cpp writes TXT, SRT, VTT, CSV or JSON. Weesper Transcribe exports to TXT, Markdown, SRT, VTT, DOCX, PDF, HTML, CSV and JSON; plain text is free, the rest is Pro.
Does Weesper Transcribe run on Intel Macs?
Yes. It requires macOS 13 or later and runs on Apple Silicon, with Metal acceleration, and on Intel Macs with AVX2 support (Haswell or later).
Turn your first MP3 into text
Converting MP3 to text on a Mac offline comes down to two routes. whisper.cpp is free and scriptable if Terminal is home; a dedicated app removes the conversion, review and export steps.
Have an MP3 waiting? See how Weesper Transcribe turns audio files into editable text, then download it from the Mac App Store: free for files up to 15 minutes, with a one-time Pro upgrade and no subscription. Setup questions are answered in the Help Center.