Every word,beautifully spoken.
Magalar turns scripts into natural, studio-grade speech in 75 languages, with subtitles for every render. One narrator or a full cast, a paragraph or a 50,000-character manuscript, a single file or a whole folder of .docx scripts.
322 voices75 languages50,000 charactersMP3 · SRT · VTT
Try it · No sign-up
Type a line. Hear it now.
No account and no card. Write up to 200 characters, choose any of 322 voices in 75 languages, and press Generate.
A few short tries per hour without an account. Demo audio is never stored.
- Up to 200 characters per try
- A few tries per hour, no account
- Audio is not stored
- Protected by Cloudflare Turnstile
- 322 neural voices
- Türkçe
- 75 languages
- Deutsch
- Dialogue studio
- 日本語
- 50,000 characters
- Français
- Word-timed subtitles
- العربية
- DOCX → ZIP
- Español
- Personal library
- हिन्दी
- SRT · VTT
- 한국어
- One seamless MP3
- Português
01Features
Every way to give text a voice.
One studio for a single narrator, a full cast, a 50,000-character manuscript or a folder full of scripts, with subtitles and a library for everything you render.
- 01
Single voice
Pick one of 322 neural voices in 75 languages and preview it. Dial in speed, pitch and volume, then generate an MP3 you can play in the browser and download.
- 322 voices
- 75 languages
- 0.5–2× speed
- ±50 Hz pitch
- 02
Dialogue studio
Build a script from blocks and give every block its own voice, even in different languages. Interviews, stories and explainers render as one continuous audio file.
- Multi-speaker
- Voice per block
- Mixed languages
- One file
- 03
Long-form
Up to 50,000 characters in a single render. Text is split at sentence boundaries, synthesized in parallel and stitched into one seamless MP3.
- 50,000 characters
- Parallel chunks
- Seamless stitch
- 04
DOCX Automator
Drop in one or many .docx scripts. Every scene is voiced and subtitled, every embedded image is extracted, and each document comes back as its own ZIP.
- Batch queue
- Scene parsing
- ZIP per document
- 05
Subtitles
Every render ships with SRT and VTT files, timed from the voice’s own word boundaries and broken into readable lines. Drop them into any editor or player.
- SRT
- VTT
- Word-timed
- Every render
- 06
Personal library
Every render is saved to your private library. Play it back, re-download the audio and subtitles, rename it or delete it. Files stay until you delete them.
- Auto-saved
- Re-download
- Rename
- Delete
02How it works
From script to sound in four steps.
No timeline to learn and nothing to install. Paste, pick, tune, generate.
03Voices
322 voices. 75 languages.
Natural neural voices across 142 locales, from Istanbul to Tokyo. Press play on an unedited sample, open any language to browse its voices, then try one in the demo.
- 01tr-TR
Emel
Turkish
Merhaba, ben Emel. Yazdığınız her metni sıcak ve doğal bir sesle sizin için okuyabilirim.
- 02tr-TR
Ahmet
Turkish
Uzun bir makaleyi ya da kitap bölümünü yapıştırın, gerisini Magalar halletsin.
- 03en-US
Aria
English, United States
- 04de-DE
Seraphina
German
Guten Tag! Ich lese Ihre Texte klar, natürlich und mit der richtigen Betonung vor.
- 05fr-FR
Vivienne
French
Bonjour ! Confiez-moi votre texte, et je vous le lirai d'une voix naturelle et posée.
- 06es-ES
Elvira
Spanish
¡Hola! Convierte cualquier texto en una narración natural, lista para tu próximo vídeo.
- 07ja-JP
Nanami
Japanese
こんにちは。どんな文章でも、自然で聞きやすい声でお読みします。
- 08ar-SA
Zariyah
Arabic
مرحبًا! أحوّل نصوصك إلى صوت طبيعي وواضح خلال ثوانٍ.
- 09hi-IN
Swara
Hindi
नमस्ते! आपका कोई भी लेख मैं साफ़ और स्वाभाविक आवाज़ में पढ़ सकती हूँ।
- 10en-GB
Ryan
English, United Kingdom
- 11en-US
Andrew
English, United States
- 12en-AU
Natasha
English, Australia
Featured
All languages A–Z
Samples are unedited renders by Magalar. Every voice can be tried in the demo or previewed in the studio.
Long-form
Chapters, not snippets.
Paste an entire chapter, lecture or report. Magalar cuts it at sentence boundaries, renders every chunk at the same time and joins them back into a single file, so there is nothing to splice by hand.
50,000
characters in a single render
- Limit
- 50,000 characters per render
- Split
- At sentence boundaries, any script
- Render
- Chunks in parallel
- Output
- One seamless MP3
04DOCX Automator
Drop in scripts. Get back finished audio.
Built for video, e-learning and slideshow production. Queue one or many .docx scripts: every scene is voiced and subtitled, every embedded image is pulled out, and each document returns as a tidy, numbered ZIP that also waits in your library.
- 01Upload .docxOne file or a batch
- 02Parse scenes“Scene N” · “Spoken script:”
- 03Voice every scene01.mp3 · 01.srt · 01.vtt
- 04Extract images01_1.png, 03_1.png …
- 05Download ZIPOne ZIP per document
module-01.zip ready · module-02.docx up next
- module-01.docxZIP ready
- module-02.docxProcessing
- module-03.docxQueued
Scene 1
Editor note: keep the intro under a minute.
Spoken script:
Welcome to module one. Over the next few minutes we will set up your workspace and record a first take.
Scene 2
Spoken script:
Start with the toolbar at the top of the screen…
Each “Scene N” starts a new scene. Only the narration after “Spoken script:” is voiced; notes and visual suggestions are skipped.
module-01.zip
- 01.mp3Scene 1 · audio
- 01.srtScene 1 · subtitles
- 01.vttScene 1 · subtitles
- 01_1.pngScene 1 · image 1
- 02.mp3Scene 2 · audio
- 02.srtScene 2 · subtitles
- 02.vttScene 2 · subtitles
- 03.mp3Scene 3 · audio
- 03_1.jpgScene 3 · image 1
- …
Audio and its SRT and VTT subtitles are numbered by scene. Images keep their scene number plus an index, so files drop straight onto a timeline or a slide deck in order.
Use cases
Made for anything that should be heard.
- 01
Video voiceovers
Narration for YouTube, explainers and product films in dozens of languages, with subtitles for every scene.
- 02
E-learning
Turn lesson scripts into consistent course audio with the DOCX Automator.
- 03
Audiobooks & long reads
Whole chapters of up to 50,000 characters, rendered as one MP3.
- 04
Podcasts & dialogue
Cast a different voice for every speaker and export a single file.
- 05
Accessibility
Offer a listenable version of articles, documentation and announcements.
- 06
Slideshows
Numbered MP3s and extracted images, ready to drop onto slides in order.
05FAQ
Questions, answered.
The short version: 75 languages, MP3 with subtitles out, up to 50,000 characters in.
01Which languages and voices are available?
322 natural neural voices across 75 languages and 142 locales: English, Turkish, German, French, Spanish, Japanese, Arabic, Hindi and many more, often with several regional accents. A handful of multilingual voices can read many languages in one consistent voice.
02Can I try it without an account?
Yes. The demo near the top of the page renders up to 200 characters with any voice, no sign-up needed. It allows a few short tries per hour per visitor, and demo audio is never stored. A free account unlocks the full studio.
03How long can a single render be?
Up to 50,000 characters. Long text is split at sentence boundaries (in any script, including Chinese, Japanese, Arabic and Hindi), the chunks are synthesized in parallel, and the results are stitched back into one seamless MP3.
04Do I get subtitles?
Yes, with every render. You get an SRT and a VTT file built from the voice’s own word timings and split into readable lines of a few seconds each. Long-form and dialogue subtitles are timed against the final stitched MP3, and the DOCX Automator adds an .srt and a .vtt for every scene.
05What is the library?
Every render is saved to your personal library automatically. Search it, play renders back, re-download the MP3, subtitles or ZIP, rename them or delete them. Only you can see your library.
06How long are my files kept?
Until you delete them. Renders are stored privately on Cloudflare R2 and delivered through private download links that expire after an hour; a fresh link is created whenever you open your library. Deleting a render removes its audio, subtitles and ZIP.
07What can I adjust?
Speed from 0.5× to 2×, pitch by up to ±50 Hz and volume by up to ±50%. You can preview voices and settings before you generate.
08What file formats do I get?
Single-voice, long-form and dialogue renders come back as an MP3 plus SRT and VTT subtitles. The DOCX Automator returns one ZIP per document, containing numbered MP3s, matching subtitle files and the extracted images.
09How should I format a .docx for the Automator?
Start each scene with a line like “Scene 1”, then put the narration after a “Spoken script:” line. Images embedded in a scene are extracted and named after it (01_1.png, 01_2.png…). Notes and visual suggestions are not voiced.
10Can I use several voices in one file?
Yes. In the dialogue studio you build a script from blocks, assign a voice to each block (in any language) and render everything as one continuous audio file with one set of subtitles.
11Do I need an account? What does it cost?
You can try the demo without one. To open the studio, create a free account with email or a social login. Magalar is free during early access.
Free during early access
Give your words a voice.
Email or social sign-in · 75 languages · No software to install