From 62bac0a1d1b38fd62fbef3ba311d462f44dfaea3 Mon Sep 17 00:00:00 2001
From: liyulingyue <852433440@qq.com>
Date: Tue, 3 Dec 2024 20:29:04 +0800
Subject: [PATCH] fix aishell3-vc0

---
 paddlespeech/t2s/modules/tacotron2/attentions.py | 4 +++-
 paddlespeech/t2s/modules/tacotron2/encoder.py    | 2 ++
 2 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/paddlespeech/t2s/modules/tacotron2/attentions.py b/paddlespeech/t2s/modules/tacotron2/attentions.py
index 5d1a24845..1676fc330 100644
--- a/paddlespeech/t2s/modules/tacotron2/attentions.py
+++ b/paddlespeech/t2s/modules/tacotron2/attentions.py
@@ -171,7 +171,9 @@ class AttLoc(nn.Layer):
         if paddle.sum(att_prev) == 0:
             # if no bias, 0 0-pad goes 0
             att_prev = 1.0 - make_pad_mask(enc_hs_len)
-            att_prev = att_prev / enc_hs_len.unsqueeze(-1)
+            unsqueeze_enc_hs_len = enc_hs_len.unsqueeze(-1).astype(
+                att_prev.dtype)
+            att_prev = att_prev / unsqueeze_enc_hs_len
 
         # att_prev: (utt, frame) -> (utt, 1, 1, frame)
         # -> (utt, att_conv_chans, 1, frame)
diff --git a/paddlespeech/t2s/modules/tacotron2/encoder.py b/paddlespeech/t2s/modules/tacotron2/encoder.py
index 224c82400..7683def83 100644
--- a/paddlespeech/t2s/modules/tacotron2/encoder.py
+++ b/paddlespeech/t2s/modules/tacotron2/encoder.py
@@ -162,6 +162,8 @@ class Encoder(nn.Layer):
             return xs.transpose([0, 2, 1])
         if not isinstance(ilens, paddle.Tensor):
             ilens = paddle.to_tensor(ilens)
+        if ilens.ndim == 0:
+            ilens = ilens.unsqueeze(0)
         xs = xs.transpose([0, 2, 1])
         # for dygraph to static graph
         # self.blstm.flatten_parameters()