print log
Browse files- vietTTS/nat/text2mel.py +6 -0
vietTTS/nat/text2mel.py
CHANGED
|
@@ -100,7 +100,13 @@ def text2mel(
|
|
| 100 |
durations = jnp.where(
|
| 101 |
np.array(tokens)[None, :] == FLAGS.word_end_index, 0.0, durations
|
| 102 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
| 103 |
mels = predict_mel(tokens, durations, acoustic_ckpt)
|
|
|
|
|
|
|
| 104 |
if tokens[-1] == FLAGS.sil_index:
|
| 105 |
end_silence = durations[0, -1].item()
|
| 106 |
silence_frame = int(end_silence * FLAGS.sample_rate / (FLAGS.n_fft // 4))
|
|
|
|
| 100 |
durations = jnp.where(
|
| 101 |
np.array(tokens)[None, :] == FLAGS.word_end_index, 0.0, durations
|
| 102 |
)
|
| 103 |
+
print("acoustic_ckpt : ")
|
| 104 |
+
print(acoustic_ckpt)
|
| 105 |
+
print("duration_ckpt : ")
|
| 106 |
+
print(duration_ckpt)
|
| 107 |
mels = predict_mel(tokens, durations, acoustic_ckpt)
|
| 108 |
+
print("mels : ")
|
| 109 |
+
print(mels)
|
| 110 |
if tokens[-1] == FLAGS.sil_index:
|
| 111 |
end_silence = durations[0, -1].item()
|
| 112 |
silence_frame = int(end_silence * FLAGS.sample_rate / (FLAGS.n_fft // 4))
|