Monitoring metric #284
Open
Monitoring metric #284
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
merge latest works
ㄴ attention alignment diagonality ㄴ average max attention weight ㄴ f0 RMSE ㄴ MCD
ㄴ re-implementing and solving errors.
ㄴ Audio processing: trimming silence(if it < 23 db), preemphasis, amplitude normalization ㄴ Remove short clip(if it < 14847 samples, It maybe percentile 0.10 my own dataset)
ㄴ Replacing MCD(metric name) to log_MCD
# Conflicts: # train.py
Closed
Yeongtae
force-pushed
the
monitoring_metric
branch
from
7b85ecf to
0f6e1b4
Compare
ㄴ debugging
Yeongtae
force-pushed
the
monitoring_metric
branch
from
0f6e1b4 to
8fad645
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary
You can see monitoring metrics for the Seq2Seq TTS model on the Tensorboard.
Monitoring metric
Attention Alignment Diagonality(AAD):
AAD is defined as the length of attention alignment path divided by the length of the diagonal path.
The attention alignment path is the line connecting maximum values for each time-step in the attention weight matrix.
The meaning of this metric is the degree of learning of the relationship between the encoder and the decoder.
Average max attention weight:
The meaning of this metric is the degree of learning of the relationship between the encoder and the decoder.
log Mel Cepstrum Distortion (MCD)
The acoustic similarity between the synthesized audio and the target audio.
f0 RMSE
The similarity of fundamental frequency between the synthesized audio and the target audio.
Future work
Example
)