From 065bf87fc5064dfecb0a3d8a65381ab7efce9ce3 Mon Sep 17 00:00:00 2001 From: khaled <khaledkoutini@gmail.com> Date: Wed, 9 Mar 2022 18:01:58 +0100 Subject: [PATCH] update to 0.0.15 --- README.md | 2 +- hear21passt/__init__.py | 2 +- hear21passt/hop100base2lvlmel.py | 2 +- hear21passt/hop160base2lvlmel.py | 2 +- 4 files changed, 4 insertions(+), 4 deletions(-) diff --git a/README.md b/README.md index b7775b8..c1eba56 100644 --- a/README.md +++ b/README.md @@ -13,7 +13,7 @@ pip3 install torch==1.8.1+cu111 torchaudio==0.8.1 -f https://download.pytorch.o # Installation Install the latest version of this repo: ```shell -pip install -e 'git+https://github.com/kkoutini/passt_hear21@0.0.14#egg=hear21passt' +pip install -e 'git+https://github.com/kkoutini/passt_hear21@0.0.15#egg=hear21passt' ``` The models follow the [common API](https://neuralaudio.ai/hear2021-holistic-evaluation-of-audio-representations.html#common-api) of HEAR 21 diff --git a/hear21passt/__init__.py b/hear21passt/__init__.py index a807122..d05ffc1 100644 --- a/hear21passt/__init__.py +++ b/hear21passt/__init__.py @@ -1,5 +1,5 @@ -__version__ = "0.0.14" +__version__ = "0.0.15" def embeding_size(hop=50, embeding_size=1000): diff --git a/hear21passt/hop100base2lvlmel.py b/hear21passt/hop100base2lvlmel.py index 79f4819..138b02a 100644 --- a/hear21passt/hop100base2lvlmel.py +++ b/hear21passt/hop100base2lvlmel.py @@ -34,7 +34,7 @@ def get_timestamp_embeddings(audio, model): """ model.eval() with torch.no_grad(): - embedmel, tmel = model.get_timestamp_mels(audio, window_size=1920) + embedmel, tmel = model.get_timestamp_mels(audio, window_size=6*100) embed1, t1 = model.get_timestamp_embeddings(audio) embed2, t2 = model.get_timestamp_embeddings(audio, window_size=model.timestamp_window * 5) # larger window embed = torch.cat((embed1, embed2, embedmel), dim=-1) diff --git a/hear21passt/hop160base2lvlmel.py b/hear21passt/hop160base2lvlmel.py index 3d2ea07..da9b7a0 100644 --- a/hear21passt/hop160base2lvlmel.py +++ b/hear21passt/hop160base2lvlmel.py @@ -34,7 +34,7 @@ def get_timestamp_embeddings(audio, model): """ model.eval() with torch.no_grad(): - embedmel, tmel = model.get_timestamp_mels(audio, window_size=1920) + embedmel, tmel = model.get_timestamp_mels(audio, window_size=6*160) # 6 windows embed1, t1 = model.get_timestamp_embeddings(audio) embed2, t2 = model.get_timestamp_embeddings(audio, window_size=model.timestamp_window * 5) # larger window embed = torch.cat((embed1, embed2, embedmel), dim=-1)