diff --git a/TTS/bin/train_glow_tts.py b/TTS/bin/train_glow_tts.py
index 9358deb2..261267df 100644
--- a/TTS/bin/train_glow_tts.py
+++ b/TTS/bin/train_glow_tts.py
@@ -182,7 +182,7 @@ def train(model, criterion, optimizer, scheduler,
         optimizer.zero_grad()
 
         # forward pass model
-        with set_amp_context(c.mixed_precision):
+        with torch.cuda.amp.autocast(enabled=c.mixed_precision):
             z, logdet, y_mean, y_log_scale, alignments, o_dur_log, o_total_dur = model.forward(
                 text_input, text_lengths, mel_input, mel_lengths, attn_mask, g=speaker_ids)
 
diff --git a/TTS/bin/train_tacotron.py b/TTS/bin/train_tacotron.py
index 09d40285..d999ddbe 100644
--- a/TTS/bin/train_tacotron.py
+++ b/TTS/bin/train_tacotron.py
@@ -158,7 +158,7 @@ def train(model, criterion, optimizer, optimizer_st, scheduler,
         if optimizer_st:
             optimizer_st.zero_grad()
 
-        with set_amp_context(c.mixed_precision):
+        with torch.cuda.amp.autocast(enabled=c.mixed_precision):
             # forward pass model
             if c.bidirectional_decoder or c.double_decoder_consistency:
                 decoder_output, postnet_output, alignments, stop_tokens, decoder_backward_output, alignments_backward = model(
diff --git a/TTS/bin/train_vocoder_wavegrad.py b/TTS/bin/train_vocoder_wavegrad.py
index 261be3fa..97eb0435 100644
--- a/TTS/bin/train_vocoder_wavegrad.py
+++ b/TTS/bin/train_vocoder_wavegrad.py
@@ -109,7 +109,7 @@ def train(model, criterion, optimizer,
 
         global_step += 1
 
-        with set_amp_context(c.mixed_precision):
+        with torch.cuda.amp.autocast(enabled=c.mixed_precision):
             # compute noisy input
             if hasattr(model, 'module'):
                 noise, x_noisy, noise_scale = model.module.compute_y_n(x)