Refactor docker dependencies to gradio

Cleanup logging
Refactor to resolve warnings
Resolve missing directory exception in audio input handling

Dockerfile

@@ -14,7 +14,7 @@ ADD . /neon_iris
 WORKDIR /neon_iris
 
 RUN pip install wheel && \
-    pip install .[docker]
+    pip install .[gradio]
 
 COPY docker_overlay/ /
 

neon_iris/client.py

@@ -298,6 +298,7 @@ def _send_serialized_message(self, serialized: dict):
                 self._connection.connection,
                 queue="neon_chat_api_request",
                 request_data=serialized)
+            LOG.debug(f"emitted {serialized.get('msg_type')}")
         except Exception as e:
             LOG.exception(e)
             self.shutdown()

neon_iris/web_client.py

@@ -23,7 +23,9 @@
 # LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
 # NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
 # SOFTWARE,  EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
-from os.path import isfile, join
+
+from os import makedirs
+from os.path import isfile, join, isdir
 from time import time
 from typing import List, Optional
 
@@ -49,6 +51,9 @@ def __init__(self, lang: str = None):
         self._await_response = Event()
         self._response = None
         self._current_tts = None
+        self._audio_path = join(xdg_data_home(), "iris", "stt")
+        if not isdir(self._audio_path):
+            makedirs(self._audio_path)
         self.lang = lang or self.config.get('default_lang') or \
                     self.config.get('languages', ['en-us'])[0]
         self.chat_ui = gradio.Blocks()
@@ -80,11 +85,12 @@ def send_audio(self, audio_file: str, lang: str = "en-us",
                    username: Optional[str] = None,
                    user_profiles: Optional[list] = None):
         """
-        @param audio_file: path to audio file to send to speech module
+        @param audio_file: path to wav audio file to send to speech module
         @param lang: language code associated with request
         @param username: username associated with request
         @param user_profiles: user profiles expecting a response
         """
+        # TODO: Audio conversion is really slow here. check ovos-stt-http-server
         audio_file = self.convert_audio(audio_file)
         self._send_audio(audio_file, lang, username, user_profiles)
 
@@ -107,7 +113,7 @@ def convert_audio(self, audio_file: str, target_sr=16000, target_channels=1,
         # Ensure the audio array is in the correct format (int16 for 2-byte samples)
         y_resampled = (y_resampled * (2 ** (8 * 2 - 1))).astype(dtype)
 
-        output_path = join(join(xdg_data_home(), "iris", "stt"), f"{time()}.wav")
+        output_path = join(self._audio_path, f"{time()}.wav")
         # Save the audio file with the new sample rate and sample width
         sf.write(output_path, y_resampled, target_sr, format='WAV', subtype='PCM_16')
         LOG.info(f"Converted audio file to {output_path}")
@@ -119,8 +125,8 @@ def on_user_input(self, utterance: str, *args, **kwargs) -> str:
         @param utterance: String utterance submitted by the user
         @returns: String response from Neon (or "ERROR")
         """
-        LOG.info(args)
-        LOG.info(kwargs)
+        # TODO: This should probably queue with a separate iterator thread
+        LOG.debug(f"args={args}|kwargs={kwargs}")
         self._await_response.clear()
         self._response = None
         if utterance:
@@ -158,8 +164,7 @@ def run(self):
             # Define primary UI
             audio_input = gradio.Audio(source="microphone",
                                        type="filepath",
-                                       label=speech,
-                                       auto_submit=True)
+                                       label=speech)
             gradio.ChatInterface(self.on_user_input,
                                  chatbot=chatbot,
                                  textbox=textbox,

requirements/gradio.txt

@@ -0,0 +1,3 @@
+gradio~=3.28
+librosa~=0.9
+soundfile~=0.12

requirements/requirements.txt

@@ -1,6 +1,6 @@
 click~=8.0
 click-default-group~=1.2
-neon-utils[audio]~=1.0
+neon-utils~=1.0
 pyyaml>=5.4,<7.0.0
 neon-mq-connector~=0.7,>=0.7.1a4
 ovos-bus-client~=0.0.3

setup.py

@@ -71,7 +71,7 @@ def get_requirements(requirements_filename: str):
     ],
     python_requires='>=3.6',
     install_requires=get_requirements("requirements.txt"),
-    extras_require={"docker": get_requirements("docker.txt")},
+    extras_require={"gradio": get_requirements("gradio.txt")},
     entry_points={
         'console_scripts': ['iris=neon_iris.cli:neon_iris_cli']
     }