| commit | 709f0ba908180130960923817558555ea72a6780 | [log] [tgz] |
|---|---|---|
| author | Viet Dang <vddang@google.com> | Thu Oct 11 17:08:30 2018 |
| committer | Viet Dang <vddang@google.com> | Fri Oct 12 13:17:52 2018 |
| tree | 02a46f03c48f96d4c69601e8dc43b333c59d1d35 | |
| parent | f21d038e0db23e416e7f63bcac3787755432f0ac [diff] |
Adds a 50-utterance test dataset for ASR model. Generated from audio utterances in the LibriSpeech ASR dataset: http://www.openslr.org/12/ How it was generated: https://docs.google.com/document/d/13DHf-xh2NdHgmeAQ2ATemiF9vCXa8ca6k9wBYA-OCtY/edit#heading=h.w379zk23nfhn The old single utterance test data was also removed. Bug: 113652468 Test: Covered by existing test: com.android.nn.benchmark.app.NNTest#testNNAPI10Seconds[asr_float] Change-Id: Ib4041236b6af33fddae5b23925498b046c1fd0a9