Merge branch 'espnet:master' into master

chintu619 · Feb 13, 2022 · 4aefb6e · 4aefb6e
2 parents eafa483 + 2ad5b51
commit 4aefb6e
Show file tree

Hide file tree

Showing 48 changed files with 322 additions and 816 deletions.
diff --git a/.gitignore b/.gitignore
@@ -43,6 +43,7 @@ egs*/*/*/wav*
 egs*/*/*/nltk*
 egs*/*/*/.cache*
 egs*/*/*/pretrained_models*
+egs*/fisher_callhome_spanish/*/local/mapping*
 
 # tools related
 tools/chainer

diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
@@ -76,6 +76,18 @@ to its differences.
 - If a recipe for a new corpus is proposed, you should add its name and information to:
 https://github.com/espnet/espnet/blob/master/egs/README.md if it's a ESPnet1 recipe,
 or https://github.com/espnet/espnet/blob/master/egs2/README.md + `db.sh` if it's a ESPnet2 recipe.
+
+#### 1.3.4 Checklist before you submit the recipe-based PR
+
+- [ ] be careful about the name for the recipe. It is recommended to follow naming conventions of the other recipes
+- [ ] common/shared files are linked with **soft link** (see Section 1.3.3)
+- [ ] modified or new python scripts should be passed through **latest** black formating (by using python package black). The command to be executed could be `black espnet espnet2 test utils setup.py egs*/*/*/local egs2/TEMPLATE/asr1/pyscripts`
+- [ ] cluster settings should be set as **default** (e.g., cmd.sh conf/slurm.conf conf/queue.conf conf/pbs.conf)
+- [ ] update `egs/README.md` or `egs2/README.md` with corresponding recipes
+- [ ] add corresponding entry in `egs2/TEMPLATE/db.sh` for a new corpus
+- [ ] try to **simplify** the model configurations. We recommend to have only the best configuration for the start of a recipe. Please also follow the default rule defined in Section 1.3.3
+- [ ] large meta-information for a corpus should be maintained elsewhere other than in the recipe itself
+- [ ] recommend to also include results and pre-trained model with the recipe
 
 ## 2 Pull Request
 If your proposed feature or bugfix is ready, please open a Pull Request (PR) at https://github.com/espnet/espnet

diff --git a/docker/build.sh b/docker/build.sh
@@ -3,6 +3,10 @@
 # 2019, Nelson Yalta
 # 2019, Ludwig Kürzinger, Technische Universität München
 
+log() {
+    local fname=${BASH_SOURCE[1]##*/}
+    echo -e "$(date '+%Y-%m-%dT%H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
+}
 
 SCRIPTPATH="$( cd "$(dirname "$0")" ; pwd -P )"
 
@@ -11,10 +15,7 @@ ubuntu_ver=20.04
 cuda_ver=11.1
 build_ver=cpu
 build_cores=24
-th_ver=1.8.0
-
-docker_ver=$(docker version -f '{{.Server.Version}}')
-echo "Using Docker Ver.${docker_ver}"
+th_ver=1.10.1
 
 
 cmd_usage() {
@@ -63,16 +64,16 @@ cmd_usage() {
 
 
 build(){
-    echo "Build Latest docker containers"
+    log "Build Latest docker containers"
     # build runtime and gpu based containers
     this_tag=espnet/espnet:runtime-latest
     docker_image=$( docker images -q  ${this_tag} )
     if ! [[ -n ${docker_image} ]]; then
-        echo "Now building Runtime container"
+        log "Now building Runtime container"
         docker build --build-arg DOCKER_VER=${docker_ver} \
                     --build-arg FROM_TAG=${default_ubuntu_ver} \
                     --build-arg NUM_BUILD_CORES=${build_cores} \
-                    -f prebuilt/runtime/Dockerfile -t ${this_tag} . | tee -a build_runtime.log > /dev/null
+                    -f prebuilt/runtime.dockerfile -t ${this_tag} . | tee -a build_runtime.log > /dev/null
 
         docker_image=$( docker images -q ${this_tag} )
         [ -z "${docker_image}" ] && exit 1
@@ -81,9 +82,9 @@ build(){
     this_tag=espnet/espnet:cuda-latest
     docker_image=$( docker images -q  ${this_tag} )
     if ! [[ -n ${docker_image} ]]; then
-        echo "Now building CUDA container"
+        log "Now building CUDA container"
         docker build --build-arg FROM_TAG=runtime-latest \
-                    -f prebuilt/devel/gpu/${default_cuda_ver}/Dockerfile -t ${this_tag} . | tee -a build_cuda.log > /dev/null
+                    -f prebuilt/gpu.dockerfile -t ${this_tag} . | tee -a build_cuda.log > /dev/null
         docker_image=$( docker images -q ${this_tag} )
         [ -z "${docker_image}" ] && exit 1
     fi
@@ -93,8 +94,11 @@ build(){
     this_tag=espnet/espnet:cpu-latest
     docker_image=$( docker images -q  ${this_tag} )
     if ! [[ -n ${docker_image} ]]; then
-        echo "Now building cpu-latest with ubuntu:${default_ubuntu_ver}"
-        docker build --build-arg FROM_TAG=runtime-latest -f prebuilt/devel/Dockerfile -t ${this_tag} . | tee -a build_cpu.log > /dev/null
+        log "Now building cpu-latest with ubuntu:${default_ubuntu_ver}"
+        docker build --build-arg FROM_TAG=runtime-latest \
+                            -f prebuilt/devel.dockerfile \
+                            --target devel \
+                            -t ${this_tag} . | tee -a build_cpu.log > /dev/null
 
         docker_image=$( docker images -q ${this_tag} )
         [ -z "${docker_image}" ] && exit 1
@@ -106,29 +110,31 @@ build(){
     this_tag=espnet/espnet:gpu-latest
     docker_image=$( docker images -q ${this_tag}  )
     if ! [[ -n ${docker_image} ]]; then
-        echo "Now building gpu-latest with ubuntu:${default_ubuntu_ver} and cuda:${default_cuda_ver}"
-        docker build ${build_args} -f prebuilt/devel/Dockerfile -t ${this_tag}  . | tee -a build_gpu.log > /dev/null
+        log "Now building gpu-latest with ubuntu:${default_ubuntu_ver} and cuda:${default_cuda_ver}"
+        docker build ${build_args} -f prebuilt/devel.dockerfile \
+                            --target devel \
+                            -t ${this_tag}  . | tee -a build_gpu.log > /dev/null
         docker_image=$( docker images -q ${this_tag} )
         [ -z "${docker_image}" ] && exit 1
     fi
 }
 
 
 build_local(){
-    echo "Building docker container: base image, and image for ${build_ver}"
+    log "Building docker container: base image, and image for ${build_ver}"
     sleep 1
 
     # prepare espnet-repo, assuming that this script is in folder espnet/docker
     cd ${SCRIPTPATH}/..
     ESPNET_ARCHIVE="./espnet-local.tar"
-    echo "Reconstructing the local repository from the last commit"
+    log "Reconstructing the local repository from the last commit"
     git archive -o docker/${ESPNET_ARCHIVE} HEAD  || exit 1
     cd ${SCRIPTPATH}
     test -r ${ESPNET_ARCHIVE} || exit 1;
     sleep 1
 
     if [ "${build_base_image}" = true ]; then
-        echo "building ESPnet base image with ubuntu:${ubuntu_ver}"
+        log "building ESPnet base image with ubuntu:${ubuntu_ver}"
         docker build --build-arg DOCKER_VER=${docker_ver} \
                     --build-arg FROM_TAG=${ubuntu_ver} \
                     --build-arg NUM_BUILD_CORES=${build_cores} \
@@ -137,11 +143,11 @@ build_local(){
     fi
 
     if [[ ${build_ver} == "cpu" ]]; then
-        echo "building ESPnet CPU Image with ubuntu:${ubuntu_ver}"
+        log "building ESPnet CPU Image with ubuntu:${ubuntu_ver}"
         docker build --build-arg FROM_TAG=runtime-local  --build-arg ESPNET_ARCHIVE=${ESPNET_ARCHIVE} \
                      -f prebuilt/local/Dockerfile -t espnet/espnet:cpu-local . || exit 1
     elif [[ ${build_ver} == "gpu" ]]; then
-        echo "building ESPnet GPU Image with ubuntu:${ubuntu_ver} and cuda:${cuda_ver}"
+        log "building ESPnet GPU Image with ubuntu:${ubuntu_ver} and cuda:${cuda_ver}"
         if [ "${build_base_image}" = true ] ; then
             docker build -f prebuilt/devel/gpu/${ver}/Dockerfile -t espnet/espnet:cuda${ver}-cudnn7 . || exit 1
         else
@@ -154,15 +160,15 @@ build_local(){
         build_args="${build_args} --build-arg ESPNET_ARCHIVE=${ESPNET_ARCHIVE}"
         docker build ${build_args} -f prebuilt/local/Dockerfile -t espnet/espnet:gpu-cuda${ver}-cudnn7-u18-local . || exit 1
     else
-        echo "Parameter invalid: " ${ver}
+        log "ERROR: Parameter invalid: " ${ver}
     fi
 
-    echo "cleanup."
+    log "cleanup."
     test -r ${ESPNET_ARCHIVE} && rm ${ESPNET_ARCHIVE}
 }
 
 run_recipe1(){
-    ./run.sh --docker-egs an4/asr1 \
+    ./run.sh --docker-egs mini_an4/asr1 \
                         --docker-cmd run.sh \
                         --docker-gpu ${1} \
                         --verbose 1 \
@@ -173,10 +179,10 @@ run_recipe1(){
 }
 
 run_recipe2(){
-   ./run.sh --docker-egs an4/asr1  \
+   ./run.sh --docker-egs mini_an4/asr1  \
                     --docker-cmd run.sh \
                     --docker-gpu ${1} \
-                    --docker-env "NLTK_DATA=/espnet/egs2/an4/asr1/nltk_data,HOME=/espnet/egs2/an4/asr1" \
+                    --docker-env "NLTK_DATA=/espnet/egs2/mini_an4/asr1/nltk_data,HOME=/espnet/egs2/mini_an4/asr1" \
                     --is-egs2 \
                     --ngpu ${2} \
                     --stage ${3} \
@@ -185,11 +191,11 @@ run_recipe2(){
 }
 
 testing(){
-    echo "Testing docker containers"
+    log "Testing docker containers"
     # Test Docker Containers with cpu setup
     run_stage=-1
     for backend in chainer pytorch; do
-        if [ -f ../egs/an4/asr1/dump/train_nodev/deltafalse/data.json ]; then 
+        if [ -f ../egs/mini_an4/asr1/dump/train_nodev/deltafalse/data.json ]; then 
             run_stage=3
         fi
         if [ ! -f .test_cpu_${backend}.done ]; then
@@ -199,7 +205,7 @@ testing(){
     done
 
     for backend in chainer pytorch; do
-        if [ -f ../egs/an4/asr1/dump/train_nodev/deltafalse/data.json ]; then 
+        if [ -f ../egs/mini_an4/asr1/dump/train_nodev/deltafalse/data.json ]; then 
             run_stage=3
         fi
         if [ ! -f .test_gpu_${backend}.done ]; then
@@ -208,7 +214,7 @@ testing(){
         fi
     done
 
-    echo "ESPnet egs Done. Press <enter> to continue with ESPnet2 egs"
+    log "ESPnet egs Done. Press <enter> to continue with ESPnet2 egs"
     read enter
     # Test for espnet2
     run_stage=-1
@@ -227,7 +233,7 @@ testing(){
 
 push(){
     for tag in runtime-latest cuda-latest cpu-latest gpu-latest;do
-        echo "docker push espnet/espnet:${tag}"
+        log "docker push espnet/espnet:${tag}"
         ( docker push espnet/espnet:${tag} )|| exit 1
     done
 }
@@ -273,14 +279,15 @@ check=true
 [ "${default_ubuntu_ver}" != "${ubuntu_ver}" ] || [ "${default_cuda_ver}" != "${cuda_ver}" ] && check=false
 
 if [ ${check} = false ] && [ "${mode}" != "fully_local" ]; then
-    echo "Error: Use of custom versions of Ubuntu (!=${default_ubuntu_ver}) and CUDA (!=${default_cuda_ver})
+    log "Error: Use of custom versions of Ubuntu (!=${default_ubuntu_ver}) and CUDA (!=${default_cuda_ver})
         is only available for <mode> == fully_local.
         Exiting... "
     exit 0;
 fi
 
+docker_ver=$(docker version -f '{{.Server.Version}}')
+log "Using Docker Ver.${docker_ver}"
 
-echo "Using Docker Ver.${docker_ver}"
 ## Application menu
 if   [[ "${mode}" == "build" ]]; then
     build
@@ -302,4 +309,4 @@ else
     cmd_usage
 fi
 
-echo "$(basename "$0") done."
+log "$(basename "$0") done."
diff --git a/docker/prebuilt/Dockerfile → docker/espnet.dockerfile b/docker/prebuilt/Dockerfile → docker/espnet.dockerfile
@@ -11,9 +11,7 @@ ARG EXTRA_LIBS
 
 RUN if [ ${EXTRA_LIBS} = true ]; then \
         cd /espnet/tools; \
-        . ./activate_python.sh; \
-        pip install parallel_wavegan; \
-        pip install git+https://github.com/cybertronai/pytorch-lamb; \
+        make extra; \
     fi
 
 # Add user to container

diff --git a/docker/prebuilt/devel/Dockerfile → docker/prebuilt/devel.dockerfile b/docker/prebuilt/devel/Dockerfile → docker/prebuilt/devel.dockerfile
@@ -1,5 +1,5 @@
 ARG FROM_TAG
-FROM espnet/espnet:${FROM_TAG}
+FROM espnet/espnet:${FROM_TAG} as devel
 LABEL maintainer "Nelson Yalta <nyalta21@gmail.com>"
 
 ARG CUDA_VER
@@ -48,11 +48,36 @@ RUN if [ -z "${CUDA_VER}" ]; then \
     . ./activate_python.sh && \
     ./installers/install_warp-ctc.sh && \
     ./installers/install_kenlm.sh && \
-    # ./installers/install_chainer_ctc.sh && \
+    ./installers/install_chainer.sh cpu && \
     conda clean --all && \
     rm -f *.tar.*  && \
     pip cache purge
 
 RUN rm -rf ../espnet
 
 WORKDIR /
+
+
+#### For local docker
+FROM devel as espnet_local
+LABEL maintainer "Nelson Yalta <nyalta21@gmail.com>"
+
+ARG CUDA_VER
+WORKDIR /
+
+# IF using a local ESPNet repository, a temporary file containing the ESPnet git repo is copied over
+ARG ESPNET_ARCHIVE=./espnet-local.tar
+COPY  ${ESPNET_ARCHIVE} /espnet-local.tar
+
+
+# Download ESPnet
+RUN echo "Getting ESPnet sources from local repository, in temporary file: " ${ESPNET_ARCHIVE}
+RUN mkdir /espnet
+RUN tar xf espnet-local.tar -C /espnet/
+RUN rm espnet-local.tar
+
+RUN cd espnet && \
+    rm -rf docker egs test utils
+
+# Install espnet
+WORKDIR /espnet/tools
diff --git a/docker/prebuilt/devel/gpu/10.0/Dockerfile b/docker/prebuilt/devel/gpu/10.0/Dockerfile