Skip to content

Conversation

@wchargin
Copy link
Contributor

Summary:
Depends on #2557 for compatibility of underlying audio functionality.
Other than that, this test just needs to execute a bunch of things in
graph mode. Parts could be converted to use more idiomatic TF 2.x style;
this is a minimal fix.

Makes progress toward #1705.

Test Plan:
Tests pass in both TF 1.x and TF 2.x, and the test file no longer uses
run_v1_only.

wchargin-branch: pea-tests-tf2x

Summary:
These tests used `tf.contrib.ffmpeg.encode_audio`, but they can actually
just use `tf.audio.encode_wav`, which has existed since TensorFlow 1.14.
This also makes the encoder itself TF 2.x–compatible.

Makes progress toward #1705.

Test Plan:
Tests pass in both TF 1.x and TF 2.x, and the `run_v1_only` decorators
have been removed:

```
$ git grep run_v1_only '*encoder*_test.py' '*audio*_test.py' | wc -l
0
```

wchargin-branch: audio-tf2x
Summary:
Depends on #2557 for compatibility of underlying audio functionality.
Other than that, this test just needs to execute a bunch of things in
graph mode. Parts could be converted to use more idiomatic TF 2.x style;
this is a minimal fix.

Makes progress toward #1705.

Test Plan:
Tests pass in both TF 1.x and TF 2.x, and the test file no longer uses
`run_v1_only`.

wchargin-branch: pea-tests-tf2x
@wchargin
Copy link
Contributor Author

Recommend reviewing with -w.

@wchargin wchargin changed the base branch from wchargin-audio-tf2x to master August 16, 2019 04:18
wchargin-source: 7ff58ebf8aa157bc14099aae2dd04f8fd4881902
wchargin-branch: pea-tests-tf2x
@wchargin wchargin merged commit ab3f37b into master Aug 16, 2019
@wchargin wchargin deleted the wchargin-pea-tests-tf2x branch August 16, 2019 05:58
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants