Recognizing Audio with Anaconda2-5.1.0-Linux-x86_64

https://medium.freecodecamp.org/how-to-use-sound-classification-with-tensorflow-on-an-iot-platform-8997eb7bbdef

docker run --name docker-audio -p 8888:8888 -v "$PWD/notebooks:/opt/notebooks" -v "$PWD/models:/root/models" -d risinsun/docker-audio-auto

docker exec -it docker-audio /bin/bash

docker run --rm -it risinsun/docker-audio-auto /bin/bash

git clone https://github.com/tensorflow/models

docker commit -p -a "Dawon Kang risinsun2@gmail.com" -m "add tensorflow models" docker-audio risinsun/docker-audio:0.1

docker push risinsun/docker-audio:0.1

http://storage.googleapis.com/asia_audioset/youtube_corpus/v1/features/features.tar.gz https://storage.googleapis.com/audioset/vggish_model.ckpt https://storage.googleapis.com/audioset/vggish_pca_params.npz

python train.py --train_data_pattern=/Volumes/Data/Work/docker-audio/models/audioset_v1_embeddings/bal_train/*.tfrecord --frame_features=True --model=FrameLevelLogisticModel --feature_names="audio_embedding" --feature_sizes="128" --batch_size="512" --num_epochs="1000" --learning_rate_decay_examples="400000" --num_classes="527" --train_dir=/Volumes/Data/Work/docker-audio/models/logs --start_new_model

python eval.py --eval_data_pattern=/Volumes/Data/Work/docker-audio/models/audioset_v1_embeddings/eval/*.tfrecord --frame_features=True --model=FrameLevelLogisticModel --feature_names="audio_embedding" --feature_sizes="128" --batch_size="512" --num_epochs="1000" --learning_rate_decay_examples="400000" --num_classes="527" --train_dir=/Volumes/Data/Work/docker-audio/models/logs --run_once=True

https://github.com/devicehive/devicehive-audio-analysis https://s3.amazonaws.com/audioanalysis/models.tar.gz

ffmpeg -i 111.mp3 -acodec pcm_s16le -ac 1 -ar 16000 out.wav