Update most profiles to 1.1 or 2.0

README.md

@@ -6,7 +6,10 @@ Speech models and supporting files for [voice2json](https://github.com/synesthes
 
 Files are contained in `<LANGUAGE>/<LOCALE>` directories. Each locale directory should contain a `SOURCE` file describing where it was sourced from. The `LICENSE` file in each locale directory covers the artifacts for that specific profile.
 
-Directories with `pocketsphinx` contain [CMU Sphinx](https://cmusphinx.github.io/) acoustic models. Directories with `kaldi` contain [Kaldi](https://kaldi-asr.org) acoustic models (either `gmm` or `nnet3`).
+* Directories with `pocketsphinx` contain [CMU Sphinx](https://cmusphinx.github.io/) acoustic models
+* Directories with `kaldi` contain [Kaldi](https://kaldi-asr.org) acoustic models (either `gmm` or `nnet3`).
+* Directories with `deepspeech` contain [Mozilla DeepSpeech](https://github.com/mozilla/DeepSpeech) acoustic models (version 0.6).
+* Directories with `julius` contain [Julius](https://github.com/julius-speech/julius) acoustic models (DNN, version 4.5).
 
 Some files are split into multiple parts so that they can be uploaded to GitHub. This is done with the `split` command:
 

bin/generate-reports.sh

@@ -29,6 +29,41 @@ if [[ -z "${profiles[*]}" ]]; then
     done < "${src_dir}/PROFILES"
 fi
 
+# -----------------------------------------------------------------------------
+# Download Profile
+# -----------------------------------------------------------------------------
+
+function download_profile {
+    profile_name="$1"
+    profile_dir="$2"
+
+    echo $profile_name
+
+    voice2json print-downloads \
+               --with-examples \
+               --no-mixed-language-model \
+               --url-format 'http://localhost:5000/{profile}/raw/master/{file}' \
+               "${profile_name}" | \
+        while read -r json; do
+            # Source URL
+            url="$(echo "${json}" | jq --raw-output .url)"
+
+            # Destination directory and file path
+            dest_file="$(echo "${json}" | jq --raw-output .file)"
+            dest_file="${profile_dir}/${dest_file}"
+
+            # Directory of destination file
+            dest_dir="$(dirname "${dest_file}")"
+
+            echo "${url} => ${dest_file}"
+            # Create destination directory and download file
+            mkdir -p "${dest_dir}"
+            curl -sSfL -o "${dest_file}" "${url}"
+        done
+}
+
+# -----------------------------------------------------------------------------
+# Generate Report
 # -----------------------------------------------------------------------------
 
 for profile in "${profiles[@]}"; do
@@ -39,19 +74,27 @@ for profile in "${profiles[@]}"; do
     rm -rf "${dest_dir}"
     mkdir -p "${dest_dir}"
 
-    echo 'Copying...'
-    cp -aR "${src_dir}/${profile}"/* "${dest_dir}/"
+    echo 'Downloading...'
+    download_profile "$(basename "${profile}")" "${dest_dir}"
 
     echo 'Training...'
     voice2json -p "${dest_dir}" --debug train-profile
 
+    # Manually copy test files
+    test_wav_dir="${src_dir}/${profile}/test"
+    if [[ -d "${test_wav_dir}" ]]; then
+        cp -aR "${test_wav_dir}" "${dest_dir}/"
+    fi
+
     closed_dir="${dest_dir}/test/closed"
     if [[ -d "${closed_dir}" ]]; then
         echo 'Testing (closed)...'
         voice2json -p "${dest_dir}" --debug test-examples --directory "${closed_dir}" | \
             jq . > "${closed_dir}/report.json"
 
         cp "${closed_dir}/report.json" "${src_dir}/${profile}/test/closed/"
+    else
+        echo "${closed_dir}" does not exist
     fi
 
     open_dir="${dest_dir}/test/open"
@@ -68,6 +111,8 @@ for profile in "${profiles[@]}"; do
             jq . > "${open_dir}/${report_name}"
 
         cp "${open_dir}/report.json" "${src_dir}/${profile}/test/open/"
+    else
+        echo "${open_dir}" does not exist
     fi
 
     echo 'Done'