Spaces:

robinhad
/

ukrainian-stt

Running

Yurii Paniv commited on Aug 30, 2020

Commit

ecc051b

•

1 Parent(s): 6452277

Add multi-language support

Files changed (6) hide show

.github/workflows/publish-docker.yml CHANGED Viewed

@@ -8,13 +8,20 @@ jobs:
     steps:
       - name: Check out the repo
         uses: actions/checkout@v2
-      - name: Download model file
         uses: dsaltares/fetch-gh-release-asset@master
         with:
           repo: "robinhad/voice-recognition-ua"
-          version: "latest"
           file: "uk.tflite"
           token: ${{ secrets.YOUR_TOKEN }}
       - name: Push to GitHub Packages
         uses: docker/build-push-action@v1
         with:

     steps:
       - name: Check out the repo
         uses: actions/checkout@v2
+      - name: Download Ukrainian model file
         uses: dsaltares/fetch-gh-release-asset@master
         with:
           repo: "robinhad/voice-recognition-ua"
+          version: "v0.1"
           file: "uk.tflite"
           token: ${{ secrets.YOUR_TOKEN }}
+      - name: Download English model file
+        uses: dsaltares/fetch-gh-release-asset@master
+        with:
+          repo: "mozilla/DeepSpeech"
+          version: "v0.7.3"
+          file: "deepspeech-0.7.3-models.tflite"
+          token: ${{ secrets.YOUR_TOKEN }}
       - name: Push to GitHub Packages
         uses: docker/build-push-action@v1
         with:

README.md CHANGED Viewed

@@ -1,6 +1,10 @@
 # voice-recognition-ua
-# make sure to download https://github.com/robinhad/voice-recognition-ua/releases/download/0.1/uk.tflite
-# before build
 ```
 export FLASK_APP=main.py
 flask run

 # voice-recognition-ua
+How to run:
+1. Make sure to download:
+2. https://github.com/robinhad/voice-recognition-ua/releases/download/0.1/uk.tflite
+3. https://github.com/mozilla/DeepSpeech/releases/download/v0.7.3/deepspeech-0.7.3-models.tflite
+How to launch:
 ```
 export FLASK_APP=main.py
 flask run

client.py CHANGED Viewed

@@ -89,10 +89,18 @@ class VersionAction(argparse.Action):
         exit(0)
-def client(audio_file):
     model_load_start = timer()
     # sphinx-doc: python_ref_model_start
-    ds = Model("./uk.tflite")
     # sphinx-doc: python_ref_model_stop
     model_load_end = timer() - model_load_start
     print('Loaded model in {:.3}s.'.format(model_load_end), file=sys.stderr)

         exit(0)
+def client(audio_file, lang="uk"):
     model_load_start = timer()
     # sphinx-doc: python_ref_model_start
+    model_path = "uk.tflite"
+    if lang not in ["en", "uk"]:
+        lang = "uk"
+    if lang == "uk":
+        model_path = "./uk.tflite"
+    if lang == "en":
+        model_path = "./deepspeech-0.7.3-models.tflite"
+    print(lang)
+    ds = Model(model_path)
     # sphinx-doc: python_ref_model_stop
     model_load_end = timer() - model_load_start
     print('Loaded model in {:.3}s.'.format(model_load_end), file=sys.stderr)

main.py CHANGED Viewed

@@ -13,10 +13,11 @@ def index():
 @app.route('/recognize', methods=["POST"])
 def recognize():
     file = request.files['file']
     audio = BytesIO()
     file.save(audio)
     audio.seek(0)
-    result = client(audio)
     return result

 @app.route('/recognize', methods=["POST"])
 def recognize():
     file = request.files['file']
+    lang = request.form["lang"]
     audio = BytesIO()
     file.save(audio)
     audio.seek(0)
+    result = client(audio, lang)
     return result

static/main.js CHANGED Viewed

@@ -8,6 +8,7 @@ var AudioContext = window.AudioContext || window.webkitAudioContext;
 var audioContext; //audio context to help us record
 const resultNode = document.getElementById('result');
 const actionButton = document.getElementById('action');
 function resultProcess(data) {
     resultNode.textContent = `Довжина тексту: ${data.length} \n
@@ -21,6 +22,7 @@ function exportWAV(blob) {
     actionButton.textContent = "Обробляється..."
     var data = new FormData()
     data.append('file', blob);
     fetch(`./recognize`, { method: "POST", body: data })
         .then(response => response.text())
         .then(resultProcess);
@@ -29,7 +31,8 @@ function record() {
     var constraints = { audio: true, video: false }
     navigator.mediaDevices.getUserMedia(constraints).then(function (stream) {
-        actionButton.textContent = "Запис..."
         actionButton.disabled = true;
         /*
             create an audio context after getUserMedia is called

 var audioContext; //audio context to help us record
 const resultNode = document.getElementById('result');
 const actionButton = document.getElementById('action');
+const langSelector = document.getElementById('lang');
 function resultProcess(data) {
     resultNode.textContent = `Довжина тексту: ${data.length} \n
     actionButton.textContent = "Обробляється..."
     var data = new FormData()
     data.append('file', blob);
+    data.append("lang", langSelector.value);
     fetch(`./recognize`, { method: "POST", body: data })
         .then(response => response.text())
         .then(resultProcess);
     var constraints = { audio: true, video: false }
     navigator.mediaDevices.getUserMedia(constraints).then(function (stream) {
+        actionButton.textContent = "Запис...";
+        resultNode.textContent = "";
         actionButton.disabled = true;
         /*
             create an audio context after getUserMedia is called

templates/hello.html CHANGED Viewed

@@ -11,11 +11,23 @@
 <body>
     <div class="container">
-        <div class="text-center">
             <h1>Демо розпізнавання української мови</h1>
             <p>Говоріть 3 секунди після натискання на кнопку, тоді отримаєте результат</p>
-            <button class="btn btn-primary" id="action" onclick="handleAction()">Почати запис (3 сек)</button>
             <div id="result"></div>
         </div>
     </div>
     <script src="https://cdn.rawgit.com/mattdiamond/Recorderjs/08e7abd9/dist/recorder.js"></script>

 <body>
     <div class="container">
+        <div class="col-12 col-md-8 col-sm-12 col-xl-6 mx-auto text-center">
             <h1>Демо розпізнавання української мови</h1>
             <p>Говоріть 3 секунди після натискання на кнопку, тоді отримаєте результат</p>
             <div id="result"></div>
+            <div class="row no-gutters">
+                <div class="col-1 col-sm-2">&nbsp;</div>
+                <div class="col-6">
+                    <button class="btn btn-primary" id="action" onclick="handleAction()">Почати запис (3 сек)</button>
+                </div>
+                <div class="col-5 col-sm-4">
+                    <select id="lang" class="browser-default custom-select">
+                        <option selected value="uk">Українська</option>
+                        <option value="en">Англійська</option>
+                    </select>
+                </div>
+            </div>
         </div>
     </div>
     <script src="https://cdn.rawgit.com/mattdiamond/Recorderjs/08e7abd9/dist/recorder.js"></script>