Add vocab

_script/print-vocabulary.sh

@@ -0,0 +1,50 @@
+#!/usr/bin/env bash
+set -e
+
+# Print out the vocabulary from Gr.fst for all zipped models in a directory.
+# Assumes fstprint is in PATH and ngramfst.so is in LD_LIBRARY_PATH.
+
+if [ -z "$2" ]; then
+    echo 'Usage: print-vocabulary <MODEL_DIR> <VOCAB_DIR>'
+    exit 1
+fi
+
+model_dir="$1"
+vocab_dir="$2"
+
+mkdir -p "${vocab_dir}"
+
+temp_dir="$(mktemp -d)"
+function finish {
+    rm -rf "${temp_dir}"
+}
+
+trap finish EXIT
+
+find "${model_dir}" -name '*.zip' -type f | \
+    while read -r zip_file; do
+        model_name="$(basename "${zip_file}" .zip)"
+        vocab_file="${vocab_dir}/${model_name}.txt"
+
+        if [ -s "${vocab_file}" ]; then
+            echo "Skipping ${model_name} (${vocab_file})"
+            continue
+        fi
+
+        model_dir="${temp_dir}/${model_name}"
+        mkdir -p "${model_dir}"
+        unzip -j "${zip_file}" "${model_name}/graph/Gr.fst" -d "${model_dir}" || \
+            unzip -j "${zip_file}" "${model_name}/Gr.fst" -d "${model_dir}" || \
+            unzip -j "${zip_file}" "${model_name}/words.txt" -d "${model_dir}" || \
+            unzip -j "${zip_file}" "${model_name}/graph/words.txt" -d "${model_dir}" || \
+            true
+
+        if [ -f "${model_dir}/words.txt" ]; then
+            cut -d' ' -f1 < "${model_dir}/words.txt" | sort | uniq > "${vocab_file}"
+        elif [ -f "${model_dir}/Gr.fst" ]; then
+            fstprint "${model_dir}/Gr.fst" | cut -f3 | sort | uniq > "${vocab_file}"
+        else
+            echo "ERROR: can't get vocabulary for ${model_name}"
+        fi
+
+    done

_vocab/.gitattributes

@@ -0,0 +1 @@
+*.txt filter=lfs diff=lfs merge=lfs -text

_vocab/vosk-model-ar-mgb2-0.4.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9322358024475930043374b723698799c47d8e99fe91d2c50d618108d0f3c354
+size 13671773

_vocab/vosk-model-br-0.8.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dcedd286aaa4b5a278ac918421684478fafe4ce2a69c8a1fe4b065c91b76295c
+size 408890

_vocab/vosk-model-en-us-0.22-lgraph.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e122967bc04600777afde3957aed0a4610e690f507861ea556334a0c48a9781
+size 2035553

_vocab/vosk-model-nl-spraakherkenning-0.6-lgraph.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d7b8b4aafab5d0679baaecc76f16fcd98e13d08fc22ab32bfbfeb693b7f67c0f
+size 1649265

_vocab/vosk-model-small-ca-0.4.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d733c70577a4283c20ef2197a1dc8b0b2d8bb1167f57ae9e67272ed75d864ba9
+size 1871468

_vocab/vosk-model-small-cn-0.22.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c9dcbb6386472cada9446707c854f3fbf3064739aac99f1e7b7b95302026d6d
+size 857113

_vocab/vosk-model-small-cs-0.4-rhasspy.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8d426881cb0c6e82f56bcf78411ac5668e64b0e04a3457ae2187a586fa935281
+size 299968

_vocab/vosk-model-small-de-0.15.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd017e93fe10248d7e291f8ca698dea3e12b51120cbf2e09468fe56d460c2bab
+size 2696357

_vocab/vosk-model-small-en-us-0.15.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0889b2edb3d5e2a6d184c6d0d511758b07df6eb7550dee1df81e1e28ec33fe77
+size 1308647

_vocab/vosk-model-small-eo-0.42.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d55aff7fdc52215418b3c0617a165beb89460030e2e4fbaeb426546cff9ef351
+size 839991

_vocab/vosk-model-small-es-0.42.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9475ff1400cf94b97561bb6c86de8b0144b4091c4deced411164fd5f758b2f3f
+size 926038

_vocab/vosk-model-small-fa-0.5.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7e2aaa51eeae0a44f9d585edab5d9c9c7d726bcb3a8451c04ffdb595516ef348
+size 6033850

_vocab/vosk-model-small-fr-0.22.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:32a0ae1bb2f38d2916e325e0d9df49164d515644bb9200d371ebde9724a87161
+size 1332818

_vocab/vosk-model-small-hi-0.22.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b5aaf7ab3c2a4d053cb984f877d24b5f9a2a52b93dcc9db47e7403a5d2f3912
+size 2845641

_vocab/vosk-model-small-it-0.22.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7843544d72e2589be787e144915d3e3d46c57f285a3810a824f726a648bece89
+size 1985053

_vocab/vosk-model-small-ja-0.22.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8231817838e979a97f7cce140e7c07faf82b6511b756ce7ec918fb6773fd3182
+size 2381760

_vocab/vosk-model-small-ko-0.22.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9555df5e8671b6ec105ef66be50f769ad9c9c07f058de37be08fee18616375f2
+size 20775443

_vocab/vosk-model-small-kz-0.15.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:04dd8fc25bad5a88f389435bdd8dc20be90edaf03016da30b76e454873755b08
+size 4381500

_vocab/vosk-model-small-nl-0.22.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f5f184771234aa534852bd6b74593190c1f18b55d5f2250ad75429b38ec68b12
+size 1034007

_vocab/vosk-model-small-pl-0.22.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7065c6e3fc661d678f351c6f932dd995698902498ba069aee13d63c231c2a752
+size 2649311

_vocab/vosk-model-small-pt-0.3.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19fe70f8d5a4793bd594eb2fb5b351577c01c72aec190ca2c22b54516db56de5
+size 958387

_vocab/vosk-model-small-ru-0.22.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c4a9500541cc728f2481515e9bac57dbf5867e60d8b6afb1c167e11d517ef2e
+size 4550334

_vocab/vosk-model-small-sv-rhasspy-0.15.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d554d4ec48bf6122d9b1e165b50641cb33f9e75b316b40f4b007b85a170bde2
+size 1003448

_vocab/vosk-model-small-tr-0.3.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95663cb9b3a6b14289c34d066f9b65a5818c17d5a99347d891e607d55e02f4e3
+size 1107458

_vocab/vosk-model-small-uk-v3-small.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02a68adb184d803ea03a3f87858fe003686d5767078b8049e4f351f08845204e
+size 12414593

_vocab/vosk-model-small-uz-0.22.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b7b73d79c57a8b1675780d76e877171eb7200c62ca32f9887c535df651f1c2d7
+size 2671687

_vocab/vosk-model-small-vn-0.4.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff6f90769d74bcae37cef627bad7fa939f14a2c409cc3b94e347c465c7fe936f
+size 128639

_vocab/vosk-model-tl-ph-generic-0.6.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b436e856c174ff360da9a53bf08de79661b6f7fe815f0db4dad7e854474e785
+size 2290976

_vocab/vosk-model-vn-0.4.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c615174246f5832d957aac9b7b0e7fb99fc4dfe9c30264bba253cc2817826430
+size 131182