SYLIN

Drama

by Fish Audio · Model · Voice & audio

promising

Top-ranked open text-to-speech model, with emotion tags and modest hardware needs

Open text-to-speech model from Fish Audio with inline meta tags to control emotion, e.g. sighs

AccessOpen weightsNeeds~8 GBDownloadweights run in about 8 GB VRAM, per user reportsSize35B parameters

Why we say so

Artificial Analysis ranks Drama 3 first among open models on its TTS leaderboard. Fish Audio says it supports inline emotion tags, and users cited by a reviewer report it running in about 8 GB of VRAM, which is a community figure, not a tested one. The license is not stated in the evidence, so check it before commercial use.

Evidence

  1. Oct 18
    Ranked number one among open models on the Artificial Analysis TTS leaderboard
  2. Oct 18
    35B total, 3B active parameters; users report it running in as little as 8 GB VRAM

Cost to run