EvalTTS

Listening reports, capability matrices, and audio samples for the open-source TTS models deployed on era-tts.

Models tested

Test cases

169 / 182

Index updated

2026-06-04

Tested models

Compare same case across models →

MOSS-TTS-Local-Transformer 1.7B

45s

4.25×

35/46

base enmultilingualcode switchingcloningpronunciationpauseslong formfailure modes

MOSS-TTS-v1.5 8B

166s

2.26×

37/37

base enmultilingualcode switchingcloningpronunciationpauseslong formfailure modes

MOSS-TTSD-v1.0 8B (dialogue)

169s

2.04×

41/41

base enmultilingualcode switchingcloningpronunciationpausesdialoguelong formfailure modes

MOSS-SoundEffect 8B

190s

1.30×

39/41

MOSS-VoiceGenerator 1.7B

139s

—

17/17

base enmultilingualcode switchingvoice designpronunciationpausesdialoguelong formfailure modes

Planned

moss-tts-realtime

MOSS-TTS · 1.7B

upstream config.json missing auto_map

moss-tts-nano

MOSS-TTS · 0.1B

Stage 8 — CPU fallback