TXT to MP3 Converter — AI Voice, 192 kbps

Text becomes an MP3 you can take anywhere

MP3 is the closest thing audio has to a universal currency. Every phone, car, laptop, smart speaker and hobbyist media player born in the last two decades plays it without complaint, every messaging and email system accepts it as an attachment, and it stays small enough to move around casually. No proprietary player, no codec hunt, no compatibility email from a confused colleague. When your words need to travel as sound, MP3 is the format that never asks questions.

That is exactly what this page produces: text-to-speech with nothing to install and no timeline to assemble. Upload a plain TXT file — meeting notes, a drafted announcement, an article draft, a study sheet — and receive an MP3 you can drop into a playlist, a chat or a podcast app. Writers hear their drafts on the commute; teachers hand students revision audio; busy people turn dead reading time into listening time. Because the output is ordinary MP3, nothing in your existing setup has to change — the file behaves like any other audio you already own. The format fades into the background, which is exactly what you want from a utility.

From text to WAV to 192 kbps MP3

The pipeline is honest about its two stages. First the Qwen3-TTS synthesis engine reads your text and renders it as speech into a WAV file — the lossless working format of the synthesis. Then FFmpeg transcodes that WAV into the finished 192 kbps MP3, which is what you download.

The 192 kbps bitrate is a deliberate choice for speech. It keeps voices crisp and free of the watery artefacts that aggressive compression adds, while shrinking the file dramatically compared with the WAV version — the MP3 is the smaller, share-friendly form of the same audio, the one you attach to a message without thinking twice. Speech also tolerates this bitrate gracefully: there are no shimmering cymbals or dense mixes to smear, so the size saving costs the listener almost nothing.

Timing is measured in minutes, not seconds: most conversions finish within a few minutes, and a long text near the input cap can take up to about 25 minutes at worst, the extra transcode step included. Progress is shown throughout; the MP3 downloads as soon as it is ready. The two-stage design is why the ceiling is stated as about 25 minutes rather than less — synthesis first, transcode second, one clock covering both.

The voice, described honestly

The speech is AI-generated by the Qwen3-TTS engine — not a recording of a human narrator, and the page does not pretend otherwise. For material that benefits from a neutral, steady read, the result is genuinely pleasant; for projects that need a performance, you will want real talent instead.

The default voice is Vivian, delivering a fixed news-broadcast style — measured, clear, professional. There are no voice pickers or speed sliders at the moment: every conversion comes out in the same composed register, which keeps results predictable from one file to the next. If a future update adds more voices, the page will say so; until then, Vivian is the voice.

Limits and privacy

Input is capped at 2000 characters — roughly nine minutes of finished audio. Texts beyond the cap are refused with a clear message rather than silently shortened. The practical habit is to split long material at chapter or section boundaries and convert the parts one at a time, in order, numbering the files as you go. Three parts of 2000 characters convert without drama; one 6000-character blob does not convert at all.

Privacy follows the platform standard. The service is free during beta with no account required, uploads travel over HTTPS, and both your TXT and the produced audio files are deleted automatically within 24 hours. Your words get a voice; they do not get a filing cabinet.