Serveurperso HF Staff bezzam HF Staff commited on
Commit
b7ee2e8
·
1 Parent(s): e0f336a

Add results from Hugging Face's Open TTS Leaderboard (#2)

Browse files

- Add results from Hugging Face's Open TTS Leaderboard (230f7cc385afd13183d8be4aac8d2fb53297dbfa)


Co-authored-by: Eric Bezzam <bezzam@users.noreply.huggingface.co>

Files changed (1) hide show
  1. README.md +8 -0
README.md CHANGED
@@ -114,6 +114,14 @@ matches the runtime target dtype on every backend. The talker LM
114
  K-quant across variants. The code predictor MTP head and the speaker
115
  encoder live in the talker GGUF and share its quantization.
116
 
 
 
 
 
 
 
 
 
117
  ## License
118
 
119
  Upstream model : Qwen3-TTS by Alibaba / Qwen team, Apache 2.0
 
114
  K-quant across variants. The code predictor MTP head and the speaker
115
  encoder live in the talker GGUF and share its quantization.
116
 
117
+ ## Evaluation
118
+
119
+ As of **September 9, 2026**, on Hugging Face's [Open TTS Leaderboard](https://huggingface.co/spaces/hf-audio/open_tts_leaderboard), Qwen3-TTS-GGUF ranks 🥇 **1st on TTFA** (time-to-first-audio) in the leaderboard's streaming benchmark — the fastest of any model to start producing audio in a single-request, streaming-mode setting.
120
+
121
+ <p align="center">
122
+ <img width="90%" alt="Top 10 models by TTFA" src="https://huggingface.co/datasets/bezzam/tts_leaderboard_screenshots/resolve/main/qwen-tts-gguf/ttfa.png" />
123
+ </p>
124
+
125
  ## License
126
 
127
  Upstream model : Qwen3-TTS by Alibaba / Qwen team, Apache 2.0