diff --git a/egs/wenetspeech4tts/TTS/f5-tts/model/cfm.py b/egs/wenetspeech4tts/TTS/f5-tts/model/cfm.py
index 349c7220e..5a30694b4 100644
--- a/egs/wenetspeech4tts/TTS/f5-tts/model/cfm.py
+++ b/egs/wenetspeech4tts/TTS/f5-tts/model/cfm.py
@@ -290,6 +290,8 @@ class CFM(nn.Module):
 
         # time step
         time = torch.rand((batch,), dtype=dtype, device=self.device)
+        # add cosyvoice cosine scheduler
+        time = 1 - torch.cos(time * 0.5 * torch.pi)
         # TODO. noise_scheduler
 
         # sample xt (φ_t(x) in the paper)
diff --git a/egs/wenetspeech4tts/TTS/f5-tts/train.py b/egs/wenetspeech4tts/TTS/f5-tts/train.py
index 35e1e8e83..4ba393af1 100755
--- a/egs/wenetspeech4tts/TTS/f5-tts/train.py
+++ b/egs/wenetspeech4tts/TTS/f5-tts/train.py
@@ -1007,13 +1007,15 @@ def run(rank, world_size, args):
         )
     else:
         raise NotImplementedError()
-
-    # scheduler = Eden(optimizer, 5000, 4, warmup_batches=params.warmup_steps)
-    warmup_scheduler = LinearLR(optimizer, start_factor=1e-8, end_factor=1.0, total_iters=params.warmup_steps)
-    decay_scheduler = LinearLR(optimizer, start_factor=1.0, end_factor=1e-8, total_iters=params.decay_steps)
-    scheduler = SequentialLR(
-        optimizer, schedulers=[warmup_scheduler, decay_scheduler], milestones=[params.warmup_steps]
-    )
+    if params.decay_steps:
+        warmup_scheduler = LinearLR(optimizer, start_factor=1e-8, end_factor=1.0, total_iters=params.warmup_steps)
+        decay_scheduler = LinearLR(optimizer, start_factor=1.0, end_factor=1e-8, total_iters=params.decay_steps)
+        scheduler = SequentialLR(
+            optimizer, schedulers=[warmup_scheduler, decay_scheduler], milestones=[params.warmup_steps]
+        )
+        assert 1==2
+    else:
+        scheduler = Eden(optimizer, 50000, 10, warmup_batches=params.warmup_steps)
 
     optimizer.zero_grad()