Is this Hebrew speech-to-text really free?

Yes. The page is public, no signup is required, and there are no length limits. The recognition model runs entirely in your browser. For higher accuracy on noisy or accented audio, run your file through Subformer's cloud transcription in Subtitles-only mode.

How does the Hebrew transcription work in the browser?

It uses an open-source neural speech recognition model, executed by a WebAssembly runtime on your own device. The model (~75 MB) is downloaded on first use and cached in your browser. Subsequent visits skip the download entirely and can transcribe offline.

Is my audio uploaded to a server?

No. Recognition runs locally in your browser. The only network calls are to download the model and the WebAssembly runtime - both are static assets. Your audio file and the generated transcript never leave your device.

What audio formats are supported?

Anything your browser can decode: MP3, WAV, M4A, AAC, FLAC, OGG/Opus, plus video containers like MP4, MOV, and WEBM (the audio track is extracted automatically).

Can I download SRT or VTT subtitles?

Yes. After transcription you can download the result as plain text (.txt), SubRip subtitles (.srt), or WebVTT (.vtt) - both subtitle formats include segment-level timestamps so you can drop them straight into any video editor or player.

Tiny vs Base - which model should I pick?

Tiny (~40 MB) is fastest and great for clean studio audio. Base (~75 MB) is roughly twice as accurate on noisy or accented audio and only slightly slower. If you're not sure, start with Base - it's the default. For the highest accuracy on hard audio (accents, noise, overlap), use Subformer's cloud transcription in Subtitles-only mode.

Does it work offline?

Once the model is cached, yes. The first visit needs network access to download it; after that the page can transcribe audio offline.

Which browsers are supported?

Any modern browser with WebAssembly: Chrome, Firefox, Safari 15+, Edge, and Chromium-based mobile browsers.

Subformer

Accede

Gratis · Sine inscriptione · In navigatore tuo currit

Gratuitus Hebrew Oratio ad Textum

עברית

Deponite fasciculum audio vel video Hebrew et accipite integrum transcriptum cum notis temporum - omnia in navigatro vestro, sine subscriptione. Modellum semel demittitur (~75 MB) et localiter in memoria reponitur, ita iterationes posteriores instantaneae sunt et sine rete operantur.

Examinatur subsidium navigatoris…

Opus est maiore accuratione vel pluribus linguis?

Nubes-fundata Hebrew Oratio ad Textum

Instrumentum liberum in navigatro currit parvum exemplar apertum - bonum pro audio mundo, sed laborare potest cum gravibus Hebrew accentibus, strepitu, vel loquentibus superimpositis, et tantum paucis linguis operit. Pro accuratione gradus emissionis, multo latiore linguae subsidio, et SRT/VTT/TXT deponendis, utere nostra soluta nube-fundata Oratione ad Textum.

Tenta nubes-fundatam Orationem ad Textum

Quomodo operat

1. Deponite fasciculum
Deponite quolibet Hebrew audio vel video - MP3, WAV, M4A, MP4, et WEBM omnia sustentantur.
2. Transcribite
Primo clic modellum ~75 MB in navigatro vestro demittit. Postea recognitio statim fit et sine rete functionem habet.
3. Demittite
Serva transcriptum ut TXT, SRT, vel VTT. Audio tuum et exitus numquam exeunt e instrumento tuo.

Audio tuum manet in apparatu tuo

Recognitio peragitur omnino in tuo navigatro utens WebAssembly. Solae petitiones retis sunt ad exemplar et ad WebAssembly runtime ad capiendum - utraque sunt res publicae, staticae. Non videmus audio tuum nec transcriptum.

Quomodo transcriptio Hebrew currit in navigatro tuo

Recognitio vocis normaliter significat imponere audio tuum ad GPU alicuius. Hic exemplar currit in machina tua propria pro, quod hoc liberum et privatum facit simul.

Open-source recognition model: Exemplar neurale apertae originis in audio multilingui exercitatum - quapropter una depositio Hebrew tractat nullo exemplari linguae proprio ad quaerendum.
WebAssembly: Tempus exsecutionis inferentiae in WebAssembly compilatur, ut codex numericus currat celeritate prope nativa intra tabulam arenam sine plugin et sine institutione.
On-device inference: Exemplar quantificatum ~75 MB oneratur et cyclus decodificationis agitur omnino in latere clientis. Nihil in pipeline ad servitorem vocat postquam pagina onerata est.
Web Audio API: Fasciculus tuus decodificatur et re-samplatur ad 16 kHz mono quod exemplar exspectat utentes propria fistula audio navigatoris, ut quodlibet formatum quod navigator tuus ludere potest accipiatur.

Premium - etiam a Subformer

Traducere et dublare

Traduc et dublā videos in plus quam 30 linguas per unum clic

Vertere subtitulos

Deponite SRT vel VTT et recipite translationem emendatam in quavis lingua.

Editor subtituli

Emenda, temporiza, et redesigne subtitulos tuos in libero editorio in navigatro.

Gratuitus Hebrew Oratio ad Textum

Nubes-fundata Hebrew Oratio ad Textum

Quomodo operat

1. Deponite fasciculum

2. Transcribite

3. Demittite

Audio tuum manet in apparatu tuo

Quomodo transcriptio Hebrew currit in navigatro tuo

Quaestiones saepe rogatae

Conare alias linguas gratuitas a voce ad textum