Merge pull request #9 from SEPIA-Framework/dev

updated README, requirements and added test script

Author: fquirin

src/README.md

@@ -2,15 +2,22 @@
 
 ### Requirements
 
-Python 3.7 is recommended.  
+Python 3.7 is recommended, 3.9 has been tested as well.  
 Please see 'requirements.txt' for more details or check out the **Dockerfile** inside the engines folder (`../engines/`).  
-Basic setup (the Vosk part might not work on all machines out-of-the-box):
+  
+Install recommended Linux packages (Debian 10|11 example):
+```
+sudo apt-get install -y python3-pip python3-dev python3-setuptools python3-wheel libffi-dev
+```
+
+Basic Pip setup (the Vosk part might not work on all machines out-of-the-box):
 
 ```
-pip install fastapi
-pip install uvicorn[standard]
-pip install aiofiles
-pip install vosk
+pip3 install cffi
+pip3 install fastapi
+pip3 install uvicorn[standard]
+pip3 install aiofiles
+pip3 install vosk
 ```
 
 ### Download the Server and ASR Models
@@ -51,4 +58,6 @@ To see all commandline options run `python -m launch --help`.
 
 ### Test
 
-Open: `http://localhost:20741/www/index.html`
+Open browser: `http://localhost:20741/www/index.html`  
+  
+Local test (Vosk): `python test_vosk.py [model-path] [test-WAV-path]`

src/requirements.txt

@@ -1,5 +1,8 @@
-fastapi ~= 0.65		# tested: 0.65.1 - newer versions should generally be no problem
-uvicorn[standard]
-aiofiles
-vosk ~= 0.3			# tested: 0.3.30 - result API changes have been announced
-# text2num ~= 2.5	# currently included custom version for German support
+# Last tested 2021.12.29 with Python 3.9.2
+#
+cffi ~= 1.15				# tested: 0.15.0
+fastapi ~= 0.70				# tested: 0.70.1
+uvicorn[standard] ~= 0.16	# tested: 0.16.0
+aiofiles ~= 0.8				# tested: 0.8.0
+vosk ~= 0.3					# tested: 0.3.32
+# text2num ~= 2.5			# custom version already included

src/test_vosk.py

@@ -0,0 +1,33 @@
+#!/usr/bin/env python3
+
+from vosk import Model, KaldiRecognizer, SetLogLevel
+import sys
+import os
+import wave
+
+SetLogLevel(0)
+
+model_path = sys.argv[1]
+if not os.path.exists(model_path):
+    print ("The model folder'", model_path, "'does not exist.")
+    exit (1)
+
+wf = wave.open(sys.argv[2], "rb")
+if wf.getnchannels() != 1 or wf.getsampwidth() != 2 or wf.getcomptype() != "NONE":
+    print ("Audio file must be WAV format mono PCM.")
+    exit (1)
+
+model = Model(model_path)
+rec = KaldiRecognizer(model, wf.getframerate())
+rec.SetWords(True)
+
+while True:
+    data = wf.readframes(4000)
+    if len(data) == 0:
+        break
+    if rec.AcceptWaveform(data):
+        print(rec.Result())
+    else:
+        print(rec.PartialResult())
+
+print(rec.FinalResult())