## [2.1] - 2021 Oct 19

### Added

- SSML support using gruut 2.0
  - <speak>, <voice name="">, <s lang="">, <w lang="" role="">, <break time="">, <say-as interpret-as="" format="">, <sub alias="">
  - voice name is <tts>:<voice>, so different TTS systems can be mixed
- Swahili, Finnish, Hungarian, Greek, and Korean voices with Glow-Speak
- Japanese, Chinese voices with Coqui-TTS
- Additional Larynx voices

### Fixed

- Single sentence synthesis with Larynx

### Removed

- Remote HTTP TTS system option

## [2.0] - 2021 Apr 29

### Added

- Larynx TTS with 40 voices (https://github.com/rhasspy/larynx)
- Embedded MaryTTS with 19 voices (https://github.com/synesthesiam/marytts-txt2wav)
- Festival Arabic voice (https://github.com/linuxscout/festival-tts-arabic-voices)
- MaryTTS-compatible endpoint at /process
- Optional WAV cache (--cache)
- New voice samples page: https://synesthesiam.github.io/opentts/

### Changed

- Docker images are split by language (synesthesiam/opentts:<LANG>) to avoid a 5GB+ image
- Allow multiple lines of text input
- Using hypercorn web server
- API test page is now at /openapi/ instead of /api/
- Dark theme for web UI

## [1.1] - 2020 Sep 29

### Added

- 2 Finnish voices for Festival
- 1 Russian voice for Festival

### Changed

- Festival Catalan/Spanish/Finnish voices use ISO-8559-15 encoding for text
- Festival Czech voices use ISO-8559-2 encoding for text
- Festival Russian voice uses Cyrillic alphabet (tranlisterated to Latin script)

## [1.0] - 2020 Jun 05

First release of OpenTTS