Skip to main content

Voice / TTS

"There are these five dangers, monks, for one who recites the Dhamma in a drawn-out singing voice." — AN 5.209

Try it: the line below is already selected — press ▶ at the bottom of the frame and listen:

5 ways to start the voice​

  1. Click the 🔉 icon next to the text's title: next to the Pali title it reads the Pali, next to a translation's title that translation.
  2. Click any Pali or translation sentence to highlight it, then press the ▶ button that appears at the bottom.
  3. Tap the ✦ before or after a line and choose Voice in its menu.
  4. Open a link to a line (for example /an5.209:1.1): the line is highlighted and ▶ is already there.
  5. Press Alt+R (Option+R on Mac).

On the search results page Alt+R reads the quotes found, in the same player: each sutta's name first, then its quotes, and the row being read is highlighted.

Player​

  • Controls — Play/Pause, previous/next sentence.
  • Navigation — pause, click a Pali sentence to highlight it, press Play to resume from there.
  • Mode — what is read: Pāḷi / Pāḷi + translation / translation / translation + Pāḷi.
  • Speed — the speed button opens a slider with presets; the default for Pāḷi is 0.7× ("normal" for the DG voice).
  • Memo — opens the Memo page.
  • AB — the A-B loop.
  • Settings (⚙) — Scroll (S: the text follows the voice), the pause between phrases in seconds, Pause only between paragraphs, and Voice settings.

Voices: three kinds​

In ⚙ → Voice settings you choose the voice separately for Pāḷi and for the translation, in three steps: kind → language → voice. The ▶ next to a voice plays a short sample (DG voices).

The first neural voices that speak Pāḷi

DG Voice are the first neural voices that read Pāḷi as plain speech, not chant: with Pāḷi's long and short vowels, and without the singing voice AN 5.209 warns about.

DG Voice — our own free neural voices (default)​

They run on our server: no key, no limits, nothing to install. Each one can also be downloaded to your device and then reads without internet.

  • Pāḷi
    • Pratham ♂ · Piper (default): reads Pāḷi by our pronunciation rules.
    • Cūḷa DG ♂ · beta: trained on real Pāḷi readings.
    • Priyamvada ♀ · Piper: a female voice, the same pronunciation rules.
  • English: Yojanāni DG ♂ (US, default), Alan ♂ (UK), Kathleen ♀ (US).
  • Russian: Cūḷa DG ♂ (default), Puriso ♂, Irina ♀.
  • Pāḷi words inside a translation (sutta, Dhamma, bhikkhu, names) are read the Pāḷi way. Abbreviations are spelled out (Ven. → Venerable, SN 56.11 → Saṁyutta Nikāya 56, 11). Russian words get their stress from the context (дела́ его / иметь де́ла), online and with a downloaded voice alike, and ё where the text prints е.
  • The next sentences are prepared while one is playing, so there are no gaps between them.

Limitations​

  • Pāḷi: rare words and long compounds are sometimes pronounced inaccurately.
  • Pāḷi: a line of one or two words sounds less clear than the same words inside a phrase.
  • Cūḷa DG (Pāḷi): the sound is still rougher than Pratham's.
  • Russian: where the stress depends on the meaning (за́мок / замо́к), it is sometimes wrong.
  • Kathleen: low sound quality.

Offline: download a DG voice​

In Voice settings → DG Voice, each voice in the list has a ⬇ icon with its download size. Once downloaded, the voice reads on your device with no network, on the reader and on the Memo page. The 🗑 icon deletes it. The English and Russian voices download a shared pronunciation module with the first of them. While a DG voice is reading online, the player itself offers Listen offline · Download with the voice's size; × hides the offer.

Cūḷa DG and Yojanāni DG voices: all rights reserved

The Cūḷa DG and Yojanāni DG voices are made from recordings of real people's voices. All rights in these voices and their models are reserved.

Permitted only: downloading these voices and listening to texts with them on the Dhamma.Gift website and in the Dhamma.Gift apps.

Prohibited without prior written permission from Dhamma.Gift:

  • any use of these voices or their models outside the Dhamma.Gift website and apps, personal and non-commercial use included;
  • copying, distributing, publishing or passing on the model files to third parties;
  • cloning or imitating these voices;
  • training or fine-tuning other models on these models or on the audio they produce;
  • any commercial use.

This restriction is also stated in the file of each such model.

Google — cloud voices with a free quota​

High-quality cloud voices. Pāḷi is read by a Punjabi (pa-IN) voice through Devanagari with our own fixes. They work with a built-in trial key; for regular use enter your own Google API key (see below), whose free monthly quota is generous for personal use.

Built-in — your device's voices, work offline​

Uses the speech synthesis of your system or browser: no internet needed, nothing is sent anywhere. There is no Pāḷi voice in any system, so install one of Sanskrit (India), Hindi (India), Nepali or Indonesian for the closest pronunciation; if none is installed, English is used as a fallback.

  • Android — Settings → Accessibility → Text-to-speech output, then install voice data for the language: Android help.
  • iPhone / iPad — Settings → Accessibility → Spoken Content → Voices: Apple help.
  • Mac — System Settings → Accessibility → Spoken Content → System voice → Manage Voices: Apple help.
  • Windows — Settings → Time & language → Speech → Manage voices → Add voices: Microsoft help.

After installing a voice, reload the page and pick it in Voice settings → Built-in.

A-B loop — repeat a passage for memorization​

The AB button next to the player is a separate tool layered on top of normal playback: pick a start and end point in the text, and the player repeats just that passage, with a pause between repeats and a repeat limit (or infinite).

A-B loop panel: A/B points, pause between repeats, repeat counter
  • Click AB — if a line was already selected, it immediately becomes point A and the panel waits for you to click a line for point B; if nothing was selected, click a line for A first, then for B. As soon as both points are set, the loop starts automatically.
  • The panel opens under the player; AB opens and closes it again.
  • Pause between loops — seconds to wait between repeats (− / + or type a number).
  • Repeats — defaults to ∞ (infinite); set your own number and the loop stops on its own once it's done. The 🗑 button clears both points.
  • To move a point, click it again and pick a new line. To clear a point, right-click (or long-press on phones) directly on the A or B button.
  • The passage between the two points is marked in the text with a vertical bar on the left.

Hotkeys​

KeyAction
Alt+RToggle Voice Mode
SToggle autoscroll
SpacePlay / Pause
← → or ↑ ↓Previous / Next segment
1Mode: Pāḷi
2Mode: Pāḷi + Translation
3Mode: Translation
4Mode: Translation + Pāḷi
+ / -Increase / decrease speed
RReset speed to default
Space and the arrow keys only work with auto-scroll on

If "Autoscroll" is off (the S key, or the toggle in the player settings), Space and the arrow keys stop responding to player control — that's expected behavior, not a bug. The 1–4 keys and +/- always work, regardless of autoscroll.

  • Voice settings — kinds of voices, languages and voices, samples, the Google API key.
  • File — a link to an audio file, when one exists for that text.
  • Help ? — this page.

Google API key (optional)​

Only for the Google voices, when the trial key isn't enough. Create a project in Google Cloud, enable the Text-to-Speech API and create an API key (how to start); paste it in Voice settings → Google API Key. Free monthly limits and prices: Google Cloud pricing. By default translations use the cheaper Standard voices, Pāḷi a premium one.

Whole texts in MP3​

For some texts, a Listen in MP3 line under the text gives the whole sutta as one file, in Pāḷi and in the translation, read by the DG voices, with the translator named. The Vinaya's two Vibhaṅgas have audio as well: Pāḷi for 347 rules and Russian for 34.

MP3 for the Memo page​

On the Memo page the ⬇ button saves the text as one MP3 in the voice you chose: with the DG voice the pauses between lines can be of any length (Google's are limited to 10 seconds).