import streamlit as st

from transcriber import Transcription

# Set app wide config

st.set_page_config(

page_title="Transcribe | Whisper UI",

page_icon="🤖",

layout="wide",

menu_items={

"Get Help": "https://twitter.com/hayabhay",

"Report a bug": "https://github.com/hayabhay/whisper-ui/issues",

"About": """This is a simple GUI for OpenAI's Whisper.

Please report any bugs or issues on [Github](https://github.com/hayabhay/whisper-ui/). Thanks!""",

},

)

# Render input type selection on the sidebar & the form

input_type = st.sidebar.selectbox("Input Type", ["YouTube", "Link", "File"])

with st.sidebar.form("input_form"):

if input_type == "YouTube":

youtube_url = st.text_input("URL (video works fine)")

if input_type == "Link":

url = st.text_input("URL (video works fine)")

elif input_type == "File":

input_file = st.file_uploader("File", type=["mp4", "avi", "mov", "mkv", "mp3", "wav"])

# Create two options. Either Youtube or a local file

name = st.text_input("Audio/Video Name", "some_cool_media")

start = st.number_input("Start time for the media (sec)", min_value=0.0, step=1.0)

duration = st.number_input("Duration (sec) - negative implies till the end", min_value=-1.0, step=1.0)

whisper_model = st.selectbox("Whisper model", options=["tiny", "base", "small", "medium", "large"], index=1)

extra_configs = st.expander("Extra Configs")

with extra_configs:

temperature = st.number_input("Temperature", min_value=0.0, max_value=1.0, value=0.0, step=0.1)

temperature_increment_on_fallback = st.number_input(

"Temperature Increment on Fallback", min_value=0.0, max_value=1.0, value=0.2, step=0.2

)

no_speech_threshold = st.slider("No Speech Threshold", min_value=0.0, max_value=1.0, value=0.6, step=0.05)

logprob_threshold = st.slider("Logprob Threshold", min_value=-20.0, max_value=0.0, value=-1.0, step=0.1)

compression_ratio_threshold = st.slider(

"Compression Ratio Threshold", min_value=0.0, max_value=10.0, value=2.4, step=0.1

)

condition_on_previous_text = st.checkbox("Condition on previous text", value=True)

transcribe = st.form_submit_button(label="Transcribe!")

if transcribe:

if not name:

st.error("Please enter a name for the audio/video")

if input_type == "YouTube":

if youtube_url and youtube_url.startswith("http"):

st.session_state.transcription = Transcription(name, youtube_url, "youtube", start, duration)

else:

st.error("Please enter a valid YouTube URL")

elif input_type == "Link":

if url and url.startswith("http"):

st.session_state.transcription = Transcription(name, url, "link", start, duration)

downloaded = True

else:

st.error("Please enter a valid URL")

elif input_type == "File":

if input_file:

st.session_state.transcription = Transcription(name, input_file, "file", start, duration)

else:

st.error("Please upload a file")

# Transcribe on click

st.session_state.transcription.transcribe(

whisper_model,

temperature,

temperature_increment_on_fallback,

no_speech_threshold,

logprob_threshold,

compression_ratio_threshold,

condition_on_previous_text,

)

# If there is a transcription, render it. If not, display instructions

if "transcription" in st.session_state:

# Render transcriptions along with the audio & video if available

transcription_col, media_col = st.columns(2, gap="large")

transcription_col.markdown("#### Trimmed audio")

with open(st.session_state.transcription.audio_path, "rb") as f:

transcription_col.audio(f.read())

transcription_col.markdown("---")

transcription_col.markdown(f"#### Transcription (whisper model - `{whisper_model}`)")

# Trim raw transcribed output off tokens to simplify

raw_output = transcription_col.expander("Raw output")

raw_output.write(st.session_state.transcription.raw_output)

transcription_col.markdown(f"##### Language: `{st.session_state.transcription.language}`")

# Show transcription in a slightly nicer format

for segment in st.session_state.transcription.segments:

transcription_col.markdown(

f"""[{round(segment["start"], 1)} - {round(segment["end"], 1)}] - {segment["text"]}"""

)

media_col.markdown("#### Original audio")

with open(st.session_state.transcription.og_audio_path, "rb") as f:

media_col.audio(f.read())

if input_type == "YouTube":

media_col.markdown("---")

media_col.markdown("#### Original YouTube Video")

media_col.video(st.session_state.transcription.source)

else:

with open("instructions.md", "r") as f:

st.write(f.read())