voice input [test mode]: init

.pre-commit-config.yaml

@@ -27,7 +27,6 @@ repos:
       - id: check-added-large-files
       - id: detect-private-key
       - id: requirements-txt-fixer
-      - id: debug-statements
 
   - repo: https://github.com/pycqa/isort
     rev: 5.12.0
@@ -52,10 +51,3 @@ repos:
       - id: python-check-blanket-noqa
       - id: python-use-type-annotations
       - id: text-unicode-replacement-char
-
-  - repo: https://github.com/pre-commit/mirrors-mypy
-    rev: v0.991
-    hooks:
-      - id: mypy
-        additional_dependencies: [ types-PyYAML==6.0.10, types_requests==2.28.8, types-ujson==5.5.0 ]
-        args: [ --ignore-missing-imports, --warn-no-return, --warn-redundant-casts, --disallow-incomplete-defs ]

.streamlit/config.toml

@@ -1,3 +1,6 @@
+[global]
+showWarningOnDirectExecution = false
+
 [theme]
 primaryColor = "#0E8388"
 backgroundColor = "#2C3333"

chat.py

@@ -10,17 +10,19 @@ from src.utils.conversation import get_user_input, show_chat_buttons, show_conve
 import streamlit as st
 
 # --- PATH SETTINGS ---
-current_dir = Path(__file__).parent if "__file__" in locals() else Path.cwd()
+current_dir: Path = Path(__file__).parent if "__file__" in locals() else Path.cwd()
-css_file = current_dir / "src/styles/.css"
+css_file: Path = current_dir / "src/styles/.css"
-assets_dir = current_dir / "assets"
+assets_dir: Path = current_dir / "assets"
-icons_dir = assets_dir / "icons"
+icons_dir: Path = assets_dir / "icons"
-img_dir = assets_dir / "img"
+img_dir: Path = assets_dir / "img"
-tg_svg = icons_dir / "tg.svg"
+tg_svg: Path = icons_dir / "tg.svg"
 
 # --- GENERAL SETTINGS ---
-PAGE_TITLE = "AI Talks"
+PAGE_TITLE: str = "AI Talks"
-PAGE_ICON = "🤖"
+PAGE_ICON: str = "🤖"
-AI_MODEL_OPTIONS = [
+LANG_EN: str = "En"
+LANG_RU: str = "Ru"
+AI_MODEL_OPTIONS: list[str] = [
     "gpt-3.5-turbo",
     "gpt-4",
     "gpt-4-32k",
@@ -35,7 +37,7 @@ with open(css_file) as f:
 
 selected_lang = option_menu(
     menu_title=None,
-    options=["En", "Ru", ],
+    options=[LANG_EN, LANG_RU, ],
     icons=["globe2", "globe"],
     menu_icon="cast",
     default_index=0,
@@ -44,36 +46,44 @@ selected_lang = option_menu(
 )
 
 # Storing The Context
+if "locale" not in st.session_state:
+    st.session_state.locale = en
 if "generated" not in st.session_state:
-    st.session_state["generated"] = []
+    st.session_state.generated = []
 if "past" not in st.session_state:
-    st.session_state["past"] = []
+    st.session_state.past = []
 if "messages" not in st.session_state:
-    st.session_state["messages"] = []
+    st.session_state.messages = []
 if "user_text" not in st.session_state:
-    st.session_state["user_text"] = ""
+    st.session_state.user_text = ""
+if "input_kind" not in st.session_state:
+    st.session_state.input_kind = st.session_state.locale.input_kind_1
 
 
 def main() -> None:
-    if st.session_state.user_text:
+    c1, c2 = st.columns(2)
-        show_conversation()
-        st.session_state.user_text = ""
-
-    c1, c2, c3 = st.columns(3)
     with c1, c2:
         c1.selectbox(label=st.session_state.locale.select_placeholder1, key="model", options=AI_MODEL_OPTIONS)
-        role_kind = c2.radio(
+        st.session_state.input_kind = c2.radio(
+            label=st.session_state.locale.input_kind,
+            options=(st.session_state.locale.input_kind_1, st.session_state.locale.input_kind_2),
+            horizontal=True,
+        )
+        role_kind = c1.radio(
             label=st.session_state.locale.radio_placeholder,
             options=(st.session_state.locale.radio_text1, st.session_state.locale.radio_text2),
             horizontal=True,
         )
         match role_kind:
             case st.session_state.locale.radio_text1:
-                c3.selectbox(label=st.session_state.locale.select_placeholder2, key="role",
+                c2.selectbox(label=st.session_state.locale.select_placeholder2, key="role",
                              options=st.session_state.locale.ai_role_options)
             case st.session_state.locale.radio_text2:
-                c3.text_input(label=st.session_state.locale.select_placeholder3, key="role")
+                c2.text_input(label=st.session_state.locale.select_placeholder3, key="role")
 
+    if st.session_state.user_text:
+        show_conversation()
+        st.session_state.user_text = ""
     get_user_input()
     show_chat_buttons()
 
@@ -85,7 +95,7 @@ if __name__ == "__main__":
         case "Ru":
             st.session_state.locale = ru
         case _:
-            locale = en
+            st.session_state.locale = en
     st.markdown(f"<h1 style='text-align: center;'>{st.session_state.locale.title}</h1>", unsafe_allow_html=True)
     st.markdown("---")
     main()

requirements.txt

@@ -1,8 +1,10 @@
-streamlit==1.20.0
+streamlit==1.21.0
 streamlit-chat==0.0.2.2
 streamlit_option_menu==0.3.2
-openai==0.27.2
+openai==0.27.4
 gtts==2.3.1
 pip==23.0.1
+bokeh==2.4.2
+streamlit-bokeh-events==0.1.2
 watchdog==3.0.0
 

src/utils/agi/bard.py

@@ -12,7 +12,7 @@ import streamlit as st
 US_AG = "Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.114 Safari/537.36"
 
 
-class Chatbot:
+class BardChat:
     """
     A class to interact with Google Bard.
     Parameters

src/utils/conversation.py

@@ -3,20 +3,34 @@ from openai.error import InvalidRequestError, OpenAIError
 from requests.exceptions import TooManyRedirects
 from streamlit_chat import message
 
-from src.utils.agi.bard import Chatbot
+from src.utils.agi.bard import BardChat
 from src.utils.agi.chat_gpt import chat_gpt_request
+from src.utils.stt import show_voice_input
 from src.utils.tts import show_player
 
 
 def clear_chat() -> None:
-    st.session_state["generated"] = []
+    st.session_state.generated = []
-    st.session_state["past"] = []
+    st.session_state.past = []
-    st.session_state["messages"] = []
+    st.session_state.messages = []
-    st.session_state["user_text"] = ""
+    st.session_state.user_text = ""
+
+
+def show_text_input() -> None:
+    st.text_area(label=st.session_state.locale.chat_placeholder, value=st.session_state.user_text, key="user_text")
 
 
 def get_user_input():
-    st.text_area(label=st.session_state.locale.chat_placeholder, value=st.session_state.user_text, key="user_text")
+    match st.session_state.input_kind:
+        case st.session_state.locale.input_kind_1:
+            clear_chat()
+            show_text_input()
+        case st.session_state.locale.input_kind_2:
+            clear_chat()
+            show_voice_input()
+            show_text_input()
+        case _:
+            show_text_input()
 
 
 def show_chat_buttons() -> None:
@@ -39,7 +53,7 @@ def show_chat(ai_content: str, user_text: str) -> None:
         st.session_state.generated.append(ai_content)
     if st.session_state.generated:
         for i in range(len(st.session_state.generated)):
-            message(st.session_state["past"][i], is_user=True, key=str(i) + "_user", avatar_style="micah")
+            message(st.session_state.past[i], is_user=True, key=str(i) + "_user", avatar_style="micah")
             message("", key=str(i))
             st.markdown(st.session_state.generated[i])
 
@@ -67,7 +81,7 @@ def chat_gpt_conversation() -> None:
 
 def bard_conversation() -> None:
     try:
-        bard = Chatbot(st.secrets.api_credentials.bard_session)
+        bard = BardChat(st.secrets.api_credentials.bard_session)
         ai_content = bard.ask(st.session_state.user_text)
         st.warning(ai_content.get("content"))
     except (TooManyRedirects, AttributeError) as err:

src/utils/lang.py

@@ -17,6 +17,10 @@ class Locale:
     chat_run_btn: str
     chat_clear_btn: str
     chat_save_btn: str
+    speak_btn: str
+    input_kind: str
+    input_kind_1: str
+    input_kind_2: str
     select_placeholder1: str
     select_placeholder2: str
     select_placeholder3: str
@@ -44,9 +48,13 @@ class EnLocale(Locale):
     donates1: str = "Russia"
     donates2: str = "World"
     chat_placeholder: str = "Start Your Conversation With AI:"
-    chat_run_btn: str = "Run"
+    chat_run_btn: str = "Ask"
     chat_clear_btn: str = "Clear"
     chat_save_btn: str = "Save"
+    speak_btn: str = "Push to Speak"
+    input_kind: str = "Input Kind"
+    input_kind_1: str = "Text"
+    input_kind_2: str = "Voice [test mode]"
     select_placeholder1: str = "Select Model"
     select_placeholder2: str = "Select Role"
     select_placeholder3: str = "Create Role"
@@ -83,9 +91,13 @@ class RuLocale(Locale):
     donates1: str = "Россия"
     donates2: str = "Остальной Мир"
     chat_placeholder: str = "Начните Вашу Беседу с ИИ:"
-    chat_run_btn: str = "Запустить"
+    chat_run_btn: str = "Спросить"
     chat_clear_btn: str = "Очистить"
     chat_save_btn: str = "Сохранить"
+    speak_btn: str = "Нажмите и Говорите"
+    input_kind: str = "Вид ввода"
+    input_kind_1: str = "Текст"
+    input_kind_2: str = "Голос [тестовый режим]"
     select_placeholder1: str = "Выберите Модель"
     select_placeholder2: str = "Выберите Роль"
     select_placeholder3: str = "Создайте Роль"

src/utils/stt.py

@@ -0,0 +1,83 @@
+import streamlit as st
+from bokeh.models import CustomJS
+from bokeh.models.widgets import Button
+from streamlit_bokeh_events import streamlit_bokeh_events
+
+REC_GIF = "assets/icons/rec_on.gif"
+
+
+def get_js_code(lang: str) -> str:
+    return """
+        var value = "";
+        var rand = 0;
+        var recognition = new webkitSpeechRecognition();
+        recognition.continuous = false;
+        recognition.interimResults = true;
+    """ + f"recognition.lang = '{lang}';" + """
+        document.dispatchEvent(new CustomEvent("GET_ONREC", {detail: 'start'}));
+
+        recognition.onspeechstart = function () {
+            document.dispatchEvent(new CustomEvent("GET_ONREC", {detail: 'running'}));
+        }
+        recognition.onsoundend = function () {
+            document.dispatchEvent(new CustomEvent("GET_ONREC", {detail: 'stop'}));
+        }
+        recognition.onresult = function (e) {
+            var value2 = "";
+            for (var i = e.resultIndex; i < e.results.length; ++i) {
+                if (e.results[i].isFinal) {
+                    value += e.results[i][0].transcript;
+                    rand = Math.random();
+
+                } else {
+                    value2 += e.results[i][0].transcript;
+                }
+            }
+            document.dispatchEvent(new CustomEvent("GET_TEXT", {detail: {t:value, s:rand}}));
+            document.dispatchEvent(new CustomEvent("GET_INTRM", {detail: value2}));
+
+        }
+        recognition.onerror = function(e) {
+            document.dispatchEvent(new CustomEvent("GET_ONREC", {detail: 'stop'}));
+        }
+        recognition.start();
+    """
+
+
+def show_speak_btn() -> Button:
+    stt_button = Button(label=st.session_state.locale.speak_btn, button_type="success", width=100)
+    stt_button.js_on_event("button_click", CustomJS(code=get_js_code(st.session_state.locale.lang_code)))
+    return stt_button
+
+
+def get_bokeh_result() -> dict:
+    stt_button = show_speak_btn()
+    return streamlit_bokeh_events(
+        bokeh_plot=stt_button,
+        events="GET_TEXT,GET_ONREC,GET_INTRM",
+        key="listen",
+        refresh_on_update=False,
+        override_height=75,
+        debounce_time=0,
+    )
+
+
+def show_voice_input() -> None:
+    if "input" not in st.session_state:
+        st.session_state.input = {"text": "", "session": 0}
+    result = get_bokeh_result()
+    if result:
+        if "GET_TEXT" in result:
+            if result.get("GET_TEXT")["t"] != "" and result.get("GET_TEXT")["s"] != st.session_state.input["session"]:
+                st.session_state.input["text"] = result.get("GET_TEXT")["t"]
+                st.session_state.input["session"] = result.get("GET_TEXT")["s"]
+        if "GET_ONREC" in result:
+            placeholder = st.container()
+            if result.get("GET_ONREC") == "start":
+                placeholder.image(REC_GIF)
+                st.session_state.input["text"] = ""
+            elif result.get("GET_ONREC") == "running":
+                placeholder.image(REC_GIF)
+            elif result.get("GET_ONREC") == "stop":
+                if st.session_state.input["text"] != "":
+                    st.session_state.user_text = st.session_state.input["text"]