remove fill_constant_batch_size_like in static model of speedyspeech, test=tts

examples/csmsc/tts3/run.sh

@@ -44,7 +44,11 @@ fi
 # paddle2onnx, please make sure the static models are in ${train_output_path}/inference first
 # we have only tested the following models so far
 if [ ${stage} -le 5 ] && [ ${stop_stage} -ge 5 ]; then
-    pip install paddle2onnx==0.9.4
+    # install paddle2onnx
+    version=$(echo `pip list |grep "paddle2onnx"` |awk -F" " '{print $2}')
+    if [[ -z "$version" || ${version} != '0.9.4' ]]; then
+        pip install paddle2onnx==0.9.4
+    fi
     ./local/paddle2onnx.sh ${train_output_path} inference inference_onnx fastspeech2_csmsc
     ./local/paddle2onnx.sh ${train_output_path} inference inference_onnx hifigan_csmsc
     ./local/paddle2onnx.sh ${train_output_path} inference inference_onnx mb_melgan_csmsc
@@ -52,6 +56,10 @@ fi
 
 # inference with onnxruntime, use fastspeech2 + hifigan by default
 if [ ${stage} -le 6 ] && [ ${stop_stage} -ge 6 ]; then
-    pip install onnxruntime
+    # install onnxruntime
+    version=$(echo `pip list |grep "onnxruntime"` |awk -F" " '{print $2}')
+    if [[ -z "$version" || ${version} != '1.10.0' ]]; then
+        pip install onnxruntime==1.10.0
+    fi
     ./local/ort_predict.sh ${train_output_path}
 fi

paddlespeech/t2s/modules/positional_encoding.py

@@ -31,8 +31,9 @@ def sinusoid_position_encoding(num_positions: int,
 
     channel = paddle.arange(0, feature_size, 2, dtype=dtype)
     index = paddle.arange(start_pos, start_pos + num_positions, 1, dtype=dtype)
-    p = (paddle.unsqueeze(index, -1) *
-         omega) / (10000.0**(channel / float(feature_size)))
+    denominator = channel / float(feature_size)
+    denominator = paddle.to_tensor([10000.0], dtype='float32')**denominator
+    p = (paddle.unsqueeze(index, -1) * omega) / denominator
     encodings = paddle.zeros([num_positions, feature_size], dtype=dtype)
     encodings[:, 0::2] = paddle.sin(p)
     encodings[:, 1::2] = paddle.cos(p)