Text to Speech & Speech to Text
Natural AI voices that read your text aloud, subtitles turned into timed audio, and accurate transcripts of any recording — free, unlimited and 100% private in your browser.
- Nothing is uploaded
- Natural AI voices
- 45+ voice languages · 99 for transcripts
- MP3 · WAV · SRT · TXT
- Free, no sign-up
Please use a current version of Chrome, Edge, Firefox or Safari (2023 or newer).
Your audio is ready
Settings changed — create the file again to apply them.
Drop your subtitle file here
or
Every line is spoken at its exact time — perfect for voice-overs, dubbing and narrated videos. You can also paste subtitles with Ctrl + V.
SRT · VTT · SBV · TXT
Processed on your device — your text and audio never leave your browser
Your voice-over is ready
Settings changed — create the file again to apply them.
Drop an audio or video file here
or
Transcribe meetings, interviews, lectures, podcasts and videos into text and subtitles. Long files are fine — it works in the background.
MP3 · WAV · M4A · OGG · FLAC · WEBM · MP4 · MOV
Processed on your device — your text and audio never leave your browser
Your transcript is ready
Settings changed — create the file again to apply them.
From text to voice — or voice to text — in three steps
No account, no upload, no waiting in a queue. The AI runs right here in your browser.
- 1
Choose what you need
Text to Speech for articles, scripts and e-learning; Subtitles to Audio for voice-overs; Speech to Text for meetings, videos and voice notes.
- 2
Pick a voice or a model
Choose from 450+ natural voices in 45+ languages, or a Whisper model for transcription. The model downloads once and is cached by your browser.
- 3
Listen and download
Press Listen to hear it instantly with read-along highlighting, then download MP3, WAV, OGG, SRT, VTT, TXT or JSON.
Everything the paid TTS sites do — without the paywall
Built for creators, students, teachers and teams in the US and Europe who care about privacy.
Truly natural voices
Kokoro and Piper neural voices sound human, with real intonation — not the robotic voices of old screen readers.
Private by design
Your text, subtitles and recordings are processed on your device. Nothing is uploaded, logged or used for training.
No limits, no sign-up
No character caps, no watermarks, no daily quotas. Convert a whole book chapter or a three-hour lecture.
Subtitles to voice-over
Each subtitle line is spoken exactly on time. Long lines are sped up naturally so the voice stays in sync with your video.
Accurate transcripts
OpenAI Whisper recognises 99 languages and exports ready-to-use SRT and VTT captions with timestamps.
Works offline
After the first download the AI model lives in your browser cache, so the tool keeps working without a connection.
Text to speech, SRT to MP3 and speech to text: a practical guide
This converter combines three tools that usually live on separate (and often paid) websites. Text to Speech turns any text into a natural voice you can download as MP3, WAV or OGG. Subtitles to Audio reads an SRT or WebVTT file line by line at the right moment, which gives you a timed voice-over for YouTube, TikTok, e-learning or dubbing. Speech to Text transcribes audio and video files — or a recording from your microphone — into editable text and subtitle files.
Unlike cloud services, the AI models run inside your browser with WebAssembly (and WebGPU when your graphics card supports it). The first time you use a voice or a transcription model, your browser downloads it once from a public model library and keeps it in its cache. Your own content is never sent anywhere, which makes the tool a safe choice for confidential documents, client interviews and school work under GDPR.
For English, the Studio voices (Kokoro) give the most natural result. For German, French, Spanish, Italian, Dutch, Polish, Portuguese and 30 more languages, choose a Piper voice marked HD or Standard. Device voices play instantly but can’t be saved as a file, because the browser does not allow recording them.
Which mode and settings to use
| Goal | Mode | Recommended settings |
|---|---|---|
| Audiobook or article to MP3 | Text to Speech | Studio / HD voice · MP3 128 kbps · 0.6 s paragraph pause |
| YouTube / TikTok voice-over | Subtitles to Audio | Speed up to fit · max 1.5× · MP3 192 kbps |
| E-learning narration | Text to Speech | Speed 0.95× · pause tags between slides · SRT export |
| Meeting or interview notes | Speech to Text | Balanced model · choose the language · TXT with timestamps |
| Captions for a video | Speech to Text | Accurate model · 42 characters per line · SRT |
| Foreign video in English | Speech to Text | Translate to English · SRT |
Tips for the best results
- Write numbers, dates and abbreviations the way they should be spoken, or add them to the pronunciation dictionary (e.g. “Dr.” → “Doctor”, “ToolCMB” → “Tool C M B”).
- Use
[pause 1s]to add breathing room between sections, list items or slides. - Click any sentence in the read-along view to listen from there; only changed sentences are generated again.
- For transcription, pick the spoken language instead of auto-detect and use the Accurate model for noisy recordings.
- Chrome and Edge on a computer with a modern graphics card can use GPU acceleration for much faster transcription.
Frequently asked questions
Everything about voices, privacy, file formats and limits.
Is this text to speech really free?
Yes. There is no sign-up, no character limit, no watermark and no paid tier. The AI runs on your own device, so it costs us nothing per conversion and we don’t need to restrict you.
Are my texts and recordings uploaded?
No. Text, subtitles and audio are processed locally in your browser. The only download is the AI model itself, which comes from a public model library and contains none of your data.
Why does the first use take longer?
The first time you use a voice or a transcription model, your browser downloads it (about 20–250 MB depending on the model). It is cached, so the next time it loads in seconds and also works offline.
Can I download the speech as MP3?
Yes — MP3, WAV and OGG are supported, mono or stereo, at the quality you choose. Text to Speech can also export an SRT subtitle file that matches the audio sentence by sentence.
Which languages are supported?
Text to speech: 45+ languages including English (US/UK), German, French, Spanish, Italian, Dutch, Polish, Portuguese, Swedish, Czech, Ukrainian, Turkish, Vietnamese and Chinese, plus the voices built into your device. Speech to text: 99 languages, with optional translation into English.
How do I turn an SRT file into audio?
Open the “Subtitles to Audio” tab, drop your .srt or .vtt file, pick a voice and press “Create voice-over”. Each line starts at its timestamp; lines that are too long are sped up naturally, and you can download an adjusted SRT if anything had to move.
How accurate is the speech to text?
It uses OpenAI’s Whisper models, which are among the most accurate open speech recognisers. “Balanced” is great for clear speech; “Accurate” handles accents, noise and technical vocabulary better. You can edit every line before exporting.
Can I use the generated voice commercially?
The models are released under permissive licences (Kokoro: Apache 2.0, Piper: MIT, Whisper: MIT), and many Piper voices allow commercial use. A few Piper voices are trained on datasets with non-commercial terms — check the voice’s model card if you publish commercially.
Does it work on iPhone and Android?
Yes. Recent Safari, Chrome and Samsung Internet run all three modes. On phones the “Fast” transcription model and Standard voices are the best balance of speed and memory.
How do I free up the storage the models use?
Open Settings → “Downloaded AI models” to see each model and its size, and remove what you no longer need.
More free, private tools
Compress, convert and edit audio, video, images and PDFs — all in your browser.