add mic input

.gitignore

@@ -1,6 +1,8 @@
 # Custom ignore
 local/
 data/
+ssl/server.crt
+ssl/server.key
 
 ### Python template
 # Byte-compiled / optimized / DLL files

Dockerfile

@@ -13,11 +13,11 @@ COPY app /app
 COPY ./requirements.txt /requirements.txt
 
 # Pip install the dependencies
-RUN pip install --upgrade pip
+RUN pip --trusted-host pypi.python.org --trusted-host files.pythonhosted.org --trusted-host pypi.org install --upgrade pip
 # For CPU only, you can use pip install git+https://github.com/MiscellaneousStuff/whisper.git
 # in place of openai-whisper.
 # Also, --extra-index-url https://download.pytorch.org/whl/cpu might be needed if you are using a CPU only machine
-RUN pip install --no-cache-dir -r /requirements.txt
+RUN pip install --trusted-host pypi.python.org --trusted-host files.pythonhosted.org --trusted-host pypi.org --no-cache-dir -r /requirements.txt
 
 # Set the working directory to /app
 WORKDIR /app
@@ -29,4 +29,8 @@ EXPOSE 8501
 VOLUME /data
 
 # Run the app
-CMD streamlit run /app/01_🏠_Home.py
+#CMD streamlit run /app/01_🏠_Home.py
+
+# Run the app on https
+COPY ssl /app/ssl
+CMD streamlit run /app/01_🏠_Home.py --server.sslCertFile '/app/ssl/server.crt'  --server.sslKeyFile '/app/ssl/server.key'

README.md

@@ -1,4 +1,4 @@
-# Streamlit UI for OpenAI's Whisper
+# Streamlit UI for OpenAI's Whisper (mic support)
 
 This is a simple [Streamlit UI](https://streamlit.io/) for [OpenAI's Whisper speech-to-text model](https://openai.com/blog/whisper/).
 It let's you download and transcribe media from YouTube videos, playlists, or local files.
@@ -25,12 +25,23 @@ conda env create -f environment.yml
 
 Note: If you're using a CPU-only machine, your runtime can be sped-up by using quantization implemented by [@MicellaneousStuff](https://github.com/MiscellaneousStuff) by swapping out `pip install openai-whisper` from `requirements.txt` and replacing it with their fork `pip install git+https://github.com/MiscellaneousStuff/whisper.git` (See related discussion here - https://github.com/hayabhay/whisper-ui/issues/20)
 
+## Setup SSL termination configuration
+
+This `mic_support` branch supports "mic input". Due to browser media access issue, you should use SSL termination configuration in a reverse proxy or load balancer, or you can use streamlit [HTTPS support](https://docs.streamlit.io/library/advanced-features/https-support). streamlit 1.20.0 or later supports HTTPS. "mic input" data will be treated as "uploaded".
+
+As the easiest way, you can create a self-signed-cert as follows:
+
+```
+cd ssl
+create-self-signed-cert.sh
+```
+
 ## Usage
 
 Once you're set up, you can run the app with:
 
 ```
-streamlit run app/01_🏠_Home.py
+streamlit run app/01_🏠_Home.py --server.sslCertFile 'ssl/server.crt'  --server.sslKeyFile 'ssl/server.key'
 ```
 
 This will open a new tab in your browser with the app. You can then select a YouTube URL or local file & click "Run Whisper" to run the model on the selected media.
@@ -43,7 +54,7 @@ Alternatively, you can run the app containerized with Docker via the included do
 docker compose up
 ```
 
-Then open up a new tab and navigate to [http://localhost:8501/](http://localhost:8501/)
+Then open up a new tab and navigate to [https://localhost:8501/](https://localhost:8501/)
 
 NOTE: For existing users, this will break the database since absolute paths of files are saved. A future fix will be added to fix this.
 

app/01_🏠_Home.py

@@ -5,6 +5,13 @@
 from config import get_page_config, get_whisper_settings, save_whisper_settings
 from core import MediaManager
 
+# for self-signed-cert, not verified but un-safe.
+import ssl
+ssl._create_default_https_context = ssl._create_unverified_context
+
+from streamlit.runtime.uploaded_file_manager import UploadedFile, UploadedFileRec
+from audiorecorder import audiorecorder
+
 st.set_page_config(**get_page_config())
 
 
@@ -40,14 +47,37 @@ def get_formatted_date(date_str: str) -> str:
 # ---------
 with st.sidebar.expander("➕   Add Media", expanded=False):
     # # Render media type selection on the sidebar & the form
-    source_type = st.radio("Media Source", ["YouTube", "Upload"], label_visibility="collapsed")
+    source_type = st.radio("Media Source", ["YouTube", "Upload", "Mic"], label_visibility="collapsed")
     with st.form("input_form"):
         if source_type == "YouTube":
             youtube_url = st.text_input("Youtube video or playlist URL")
         elif source_type == "Upload":
             input_files = st.file_uploader(
                 "Add one or more files", type=["mp4", "avi", "mov", "mkv", "mp3", "wav"], accept_multiple_files=True
             )
+        elif source_type == "Mic":
+            audio = audiorecorder("Click to record", "Recording...")
+            if len(audio) > 0:
+                # To play audio in frontend:
+                #st.audio(audio.tobytes())
+
+                # To save audio to a file:
+                #wav_file = open("audio.mp3", "wb")
+                #wav_file.write(audio.tobytes())
+                #wav_file.close()
+
+                # emulate "uploaded file"
+                now = datetime.now()
+                d = f'{now:%Y%m%d%H%M%S}'
+
+                filerec = UploadedFileRec(0,
+                                          f"mic_input_{d}.mp3",
+                                          "mp3",
+                                          audio.tobytes())
+                input_files = UploadedFile(filerec)
+            else:
+                input_files = None
+
         task_options = ["transcribe", "translate"]
         task = st.selectbox(
             "Task", options=task_options, index=task_options.index(st.session_state.whisper_params["task"])
@@ -66,6 +96,11 @@ def get_formatted_date(date_str: str) -> str:
                 source = input_files
             else:
                 st.error("Please upload files")
+        elif source_type == "Mic":
+            if input_files:
+                source = input_files
+            else:
+                st.error("Please input your voice via mic") 
 
         # Lowercase the source type
         source_type = source_type.lower()
@@ -156,7 +191,7 @@ def get_formatted_date(date_str: str) -> str:
                 # Add a meta caption
                 st.write(f"#### {media['source_name']}")
 
-                source_type = "YouTube" if media["source_type"] == "youtube" else "upload"
+                source_type = "YouTube" if media["source_type"] == "youtube" else media["source_type"]
                 st.markdown(
                     f"""
                     <i>Source</i>: {source_type}<br/>
@@ -179,7 +214,7 @@ def get_formatted_date(date_str: str) -> str:
                 # Render the media
                 if media["source_type"] == "youtube":
                     st.video(media["source_link"])
-                elif media["source_type"] == "upload":
+                elif media["source_type"] in ["upload", "mic"]:
                     st.audio(media["filepath"])
 
             st.write("---")
@@ -207,6 +242,12 @@ def get_formatted_date(date_str: str) -> str:
 
                     # Add a meta caption
                     source_type = "YouTube" if media["source_type"] == "youtube" else "uploaded"
+                    if media["source_type"] == "youtube":
+                        source_type = "YouTube"
+                    elif media["source_type"] == "upload":
+                        source_type = "uploaded"
+                    else:
+                        source_tpye = media["source_type"]
                     st.markdown(
                         f"""
                         <i>Source</i>: <b>{media['source_name']}</b> ({source_type})<br/>
@@ -253,14 +294,27 @@ def get_formatted_date(date_str: str) -> str:
     if media["source_type"] == "youtube":
         st.sidebar.audio(media["filepath"], start_time=st.session_state.selected_media_offset)
         st.sidebar.video(media["source_link"])
-    elif media["source_type"] == "upload":
+    elif media["source_type"] in ["upload", "mic"]:
         st.sidebar.audio(media["filepath"], start_time=st.session_state.selected_media_offset)
 
     st.write(f'## {media["source_name"]}')
 
+    with st.expander("🔽 &nbsp; Downloads"):
+        fpath = str(Path(media["filepath"]).parent)
+        fname = str(Path(media["filepath"]).parent.name)
+        for ext in (".srt",".vtt",".json",".txt",".tsv"):
+            with open(fpath + '/transcript' + ext) as f:
+                st.download_button('Download ' + ext, f, file_name=fname+ext);
+
     with st.expander("📝 &nbsp; Metadata"):
         # Add a meta caption
         source_type = "YouTube" if media["source_type"] == "youtube" else "uploaded"
+        if media["source_type"] == "youtube":
+            source_type = "YouTube"
+        elif media["source_type"] == "upload":
+            source_type = "uploaded"
+        else:
+            source_tpye = media["source_type"]
         st.markdown(
             f"""
             <i>Source</i>: <b>{media['source_name']}</b> ({source_type})<br/>

app/core.py

@@ -117,7 +117,7 @@ def _create(self, source_list: List[Union[str, Any]], source_type: str, whisper_
                     save_dir = self.media_dir / f"""{source_dirname}-{datetime.now().strftime("%Y-%m-%d %H-%M-%S")}"""
                 # Download the audio file
                 yc.streams.get_by_itag(140).download(save_dir, filename=save_filename)
-            elif source_type == "upload":
+            elif source_type in ["upload", "mic"]:
                 # Parse the file name from the source
                 tokens = source.name.split(".")
                 source_name = ".".join(tokens[:-1])
@@ -162,7 +162,7 @@ def add(self, source: Union[str, Any], source_type: str, **whisper_args):
                 source_list = list(Playlist(source))
             else:
                 source_list = [source]
-        elif source_type == "upload":
+        elif source_type in ["upload", "mic"]:
             source_list = source if type(source) == list else [source]
 
         # Download & save the audio files and add them to the database

app/pages/02_⚙️_Settings.py

@@ -14,7 +14,7 @@
 # --------------
 st.write("### ⚙️ Whisper Settings")
 with st.form("whisper_settings_form"):
-    model_options = ["tiny", "base", "small", "medium", "large"]
+    model_options = ["tiny", "base", "small", "medium", "large", "large-v2"]
     selected_model = model_options.index(st.session_state.whisper_params["whisper_model"])
     whisper_model = st.selectbox("Model", options=model_options, index=selected_model)
     temperature = st.number_input(

docker-compose.yml

@@ -11,5 +11,6 @@ services:
         # TODO: Fix this to relative paths from some user-defined directory
       - ./data:/data
     restart: unless-stopped
+    init: true  # https://qiita.com/sun33/items/d62744f5746f6815f3c8
     ports:
       - 8501:8501

environment.yml

@@ -75,7 +75,8 @@ dependencies:
     - six==1.16.0
     - smmap==5.0.0
     - sqlalchemy==2.0.2
-    - streamlit==1.17.0
+    - streamlit==1.20.0
+    - streamlit-audiorecorder==0.0.2
     - tokenizers==0.13.2
     - toml==0.10.2
     - toolz==0.12.0

requirements.txt

@@ -11,4 +11,8 @@ SQLAlchemy==2.0.3 # https://github.com/sqlalchemy/sqlalchemy
 
 # Frontend
 # --------------------
-streamlit==1.18.1 # https://github.com/streamlit/streamlit
+streamlit==1.20.0 # https://github.com/streamlit/streamlit
+
+# Audio Recorder on Frontend
+# --------------------
+streamlit-audiorecorder==0.0.2 # https://github.com/theevann/streamlit-audiorecorder

ssl/create-self-signed-cert.sh

@@ -0,0 +1,12 @@
+#!/bin/bash
+
+#openssl genrsa 2048 > server.key
+#openssl req -new -key server.key > server.csr
+#openssl x509 -days 3650 -req -sha256 -signkey server.key < server.csr > server.crt
+#openssl x509 -text < server.crt
+
+# https://qiita.com/miyuki_samitani/items/b19aa5ac3b3c6e312bd5
+# https://qiita.com/sanyamarseille/items/46fc6ff5a0aca12e1946
+
+# https://www.karakaram.com/creating-self-signed-certificate/
+openssl req -x509 -sha256 -nodes -days 3650 -newkey rsa:2048 -subj /CN=localhost -keyout server.key -out server.crt