diff --git a/src/f5_tts/infer/utils_infer.py b/src/f5_tts/infer/utils_infer.py
index f451efc..33f9a3f 100644
--- a/src/f5_tts/infer/utils_infer.py
+++ b/src/f5_tts/infer/utils_infer.py
@@ -119,12 +119,16 @@ def load_vocoder(vocoder_name="vocos", is_local=False, local_path="", device=dev
 asr_pipe = None
 
 
-def initialize_asr_pipeline(device=device):
+def initialize_asr_pipeline(device=device, dtype=None):
+    if dtype is None:
+        dtype = (
+            torch.float16 if device == "cuda" and torch.cuda.get_device_properties(device).major >= 6 else torch.float32
+        )
     global asr_pipe
     asr_pipe = pipeline(
         "automatic-speech-recognition",
         model="openai/whisper-large-v3-turbo",
-        torch_dtype=torch.float16,
+        torch_dtype=dtype,
         device=device,
     )
 
diff --git a/src/f5_tts/model/trainer.py b/src/f5_tts/model/trainer.py
index feb1735..bd96da7 100644
--- a/src/f5_tts/model/trainer.py
+++ b/src/f5_tts/model/trainer.py
@@ -325,7 +325,9 @@ class Trainer:
 
                     if self.log_samples and self.accelerator.is_local_main_process:
                         ref_audio, ref_audio_len = vocoder.decode(batch["mel"][0].unsqueeze(0)), mel_lengths[0]
-                        torchaudio.save(f"{log_samples_path}/step_{global_step}_ref.wav", ref_audio, target_sample_rate)
+                        torchaudio.save(
+                            f"{log_samples_path}/step_{global_step}_ref.wav", ref_audio.cpu(), target_sample_rate
+                        )
                         with torch.inference_mode():
                             generated, _ = self.accelerator.unwrap_model(self.model).sample(
                                 cond=mel_spec[0][:ref_audio_len].unsqueeze(0),
@@ -336,8 +338,12 @@ class Trainer:
                                 sway_sampling_coef=sway_sampling_coef,
                             )
                         generated = generated.to(torch.float32)
-                        gen_audio = vocoder.decode(generated[:, ref_audio_len:, :].permute(0, 2, 1).cpu())
-                        torchaudio.save(f"{log_samples_path}/step_{global_step}_gen.wav", gen_audio, target_sample_rate)
+                        gen_audio = vocoder.decode(
+                            generated[:, ref_audio_len:, :].permute(0, 2, 1).to(self.accelerator.device)
+                        )
+                        torchaudio.save(
+                            f"{log_samples_path}/step_{global_step}_gen.wav", gen_audio.cpu(), target_sample_rate
+                        )
 
                 if global_step % self.last_per_steps == 0:
                     self.save_checkpoint(global_step, last=True)