Add new azure speech api support

.gitignore

@@ -81,3 +81,6 @@ proxies.txt
 # speech-to-text module folders
 deepspeech
 pocketsphinx
+
+# visual studio code editor configs
+.vscode

examples/nonocaptcha.example.yaml

@@ -15,6 +15,10 @@ speech:
         model_dir: pocketsphinx/model
     azure:
         api_subkey:
+    azurespeech:
+        region:
+        subscription_key:
+        language_type:
     amazon:
         secret_key_id:
         secret_access_key:

nonocaptcha/audio.py

@@ -11,7 +11,7 @@
 from aiohttp.client_exceptions import ClientError
 
 from nonocaptcha import util
-from nonocaptcha.speech import Amazon, Azure, Sphinx, DeepSpeech
+from nonocaptcha.speech import Amazon, Azure, Sphinx, DeepSpeech, AzureSpeech
 from nonocaptcha.base import Base
 from nonocaptcha.exceptions import DownloadError, ReloadError, TryAgain
 
@@ -27,7 +27,7 @@ def __init__(self, page, loop, proxy, proxy_auth, proc_id):
     async def solve_by_audio(self):
         """Go through procedures to solve audio"""
         await self.get_frames()
-        for i in range(10):
+        for _ in range(10):
             try:
                 answer = await self.loop.create_task(self.get_audio_response())
             except DownloadError:
@@ -73,8 +73,15 @@ async def get_audio_response(self):
         else:
             answer = None
             service = self.speech_service.lower()
-            if service in ["azure", "pocketsphinx", "deepspeech"]:
-                if service == "azure":
+            if service in [
+                "azure",
+                "pocketsphinx",
+                "deepspeech",
+                "azurespeech"
+            ]:
+                if service == "azurespeech":
+                    speech = AzureSpeech()
+                elif service == "azure":
                     speech = Azure()
                 elif service == "pocketsphinx":
                     speech = Sphinx()

nonocaptcha/nonocaptcha.example.yaml

@@ -17,6 +17,10 @@ speech:
         model_dir: pocketsphinx/model
     azure:
         api_subkey:
+    azurespeech:
+        region:
+        subscription_key:
+        language_type:
     amazon:
         secret_key_id:
         secret_access_key:

nonocaptcha/speech.py

@@ -13,6 +13,7 @@
 import sys
 import time
 import websockets
+import requests
 
 from datetime import datetime
 from uuid import uuid4
@@ -178,6 +179,63 @@ async def get_text(self, audio_data):
         await transcribe._endpoint._aio_session.close()
 
 
+class AzureSpeech(object):
+    API_REGION = settings["speech"]["azurespeech"]["region"]
+    SUB_KEY = settings["speech"]["azurespeech"]["subscription_key"]
+    language_type = settings["speech"]["azurespeech"]['language_type']
+
+    @util.threaded
+    def extract_json_body(self, response):
+        return json.loads(response)
+
+    async def bytes_from_file(self, filename):
+        async with aiofiles.open(filename, "rb") as f:
+            chunk = await f.read()
+            return chunk
+
+    async def get_text(self, mp3_filename):
+        ''' return text result or None '''
+        # convert mp3 file to WAV
+        wav_filename = await mp3_to_wav(mp3_filename)
+        # read bytes from WAV file.
+        wav_bytes = await self.bytes_from_file(wav_filename)
+        # get result of speech
+        headers = {
+            'Ocp-Apim-Subscription-Key': self.SUB_KEY,
+            'Accept': 'application/json;text/xml',
+            'Content-Type': 'audio/wav; codecs=audio/pcm; samplerate=16000',
+        }
+
+        speech_to_text_url = (
+            f"https://{self.API_REGION}.stt.speech.microsoft.com/speech/"
+            f"recognition/conversation/cognitiveservices/v1?"
+            f"language={self.language_type}&format=detailed"
+        )
+
+        response = requests.post(
+            speech_to_text_url,
+            headers=headers,
+            data=wav_bytes
+        )
+        if response.status_code == 200:
+            print(response.content)
+            content = await self.extract_json_body(response.content)
+            if (
+                "RecognitionStatus" in content
+                and content["RecognitionStatus"] == "Success"
+            ):
+                answer = content["NBest"][0]["Lexical"]
+                return answer
+            if (
+                "RecognitionStatus" in content
+                and content["RecognitionStatus"] == "EndOfDictation"
+            ):
+                return
+        else:
+            print(response.status_code)
+            return None
+
+
 class Azure(object):
     SUB_KEY = settings["speech"]["azure"]["api_subkey"]