Update app.py

app.py

@@ -1,79 +1,68 @@
-pip install soundfile
-import torch
-from transformers import pipeline
 import streamlit as st
-import io
-import soundfile as sf
+import cv2
+import numpy as np
+import os
 
-MODEL_NAME = "JackismyShephard/whisper-tiny-finetuned-minds14"
-BATCH_SIZE = 8
+def images_to_video(image_folder_path, output_video_path, fps):
+    # Get the list of image files
+    image_files = sorted(os.listdir(image_folder_path))
+    image_files = [f for f in image_files if f.endswith(('.jpg', '.jpeg', '.png'))]
 
-device = 0 if torch.cuda.is_available() else "cpu"
+    # Read the first image to get dimensions
+    first_image = cv2.imread(os.path.join(image_folder_path, image_files[0]))
+    height, width, _ = first_image.shape
 
-pipe = pipeline(
-    task="automatic-speech-recognition",
-    model=MODEL_NAME,
-    chunk_length_s=30,
-    device=device,
-)
+    # Create video writer object
+    fourcc = cv2.VideoWriter_fourcc(*'mp4v')
+    video_writer = cv2.VideoWriter(output_video_path, fourcc, fps, (width, height))
 
+    # Loop through images and write to video
+    for image_file in image_files:
+        image_path = os.path.join(image_folder_path, image_file)
+        frame = cv2.imread(image_path)
+        video_writer.write(frame)
 
-# Copied from https://github.com/openai/whisper/blob/c09a7ae299c4c34c5839a76380ae407e7d785914/whisper/utils.py#L50
-def format_timestamp(
-    seconds: float, always_include_hours: bool = False, decimal_marker: str = "."
-):
-    if seconds is not None:
-        milliseconds = round(seconds * 1000.0)
+    # Release video writer
+    video_writer.release()
 
-        hours = milliseconds // 3_600_000
-        milliseconds -= hours * 3_600_000
-
-        minutes = milliseconds // 60_000
-        milliseconds -= minutes * 60_000
-
-        seconds = milliseconds // 1_000
-        milliseconds -= seconds * 1_000
-
-        hours_marker = f"{hours:02d}:" if always_include_hours or hours > 0 else ""
-        return f"{hours_marker}{minutes:02d}:{seconds:02d}{decimal_marker}{milliseconds:03d}"
-    else:
-        # we have a malformed timestamp so just return it as is
-        return seconds
+def main():
+    st.title("Image to Video Converter")
 
+    # File uploader for selecting images
+    st.write("Please upload the images you want to convert to a video:")
+    uploaded_files = st.file_uploader("Upload Images", accept_multiple_files=True, type=["jpg", "jpeg", "png"])
 
-def transcribe(file, return_timestamps):
-    # Read audio data from the file object
-    audio_data, sample_rate = sf.read(file)
+    if uploaded_files:
+        st.write("You've uploaded the following images:")
+        for uploaded_file in uploaded_files:
+            st.image(uploaded_file)
 
-    outputs = pipe(
-        audio_data,
-        sampling_rate=sample_rate,
-        return_timestamps=return_timestamps,
-    )
-    text = outputs["text"]
-    if return_timestamps:
-        timestamps = outputs["chunks"]
-        timestamps = [
-            f"[{format_timestamp(chunk['timestamp'][0])} -> {format_timestamp(chunk['timestamp'][1])}] {chunk['text']}"
-            for chunk in timestamps
-        ]
-        text = "\n".join(str(feature) for feature in timestamps)
-    return text
+        # Parameters for video creation
+        fps = st.number_input("Frames per Second (FPS)", value=24, min_value=1)
+        output_video_name = st.text_input("Output Video Name (include .mp4)", "output.mp4")
 
+        # Convert images to video
+        if st.button("Convert to Video"):
+            temp_folder = "./temp_images"
+            os.makedirs(temp_folder, exist_ok=True)
 
-def main():
-    st.title("Automatic Speech Recognition")
+            # Save uploaded images to a temporary folder
+            for i, uploaded_file in enumerate(uploaded_files):
+                file_path = os.path.join(temp_folder, f"image_{i}.png")
+                with open(file_path, "wb") as f:
+                    f.write(uploaded_file.getbuffer())
 
-    uploaded_file = st.file_uploader("Choose an audio file...", type=["wav", "mp3", "flac"])
-    return_timestamps = st.checkbox("Return Timestamps")
+            # Convert images to video
+            images_to_video(temp_folder, output_video_name, fps)
 
-    if uploaded_file is not None:
-        st.audio(uploaded_file, format='audio')
+            # Display download link for the video
+            st.markdown(f"Download the video [here](./{output_video_name})")
 
-        if st.button("Transcribe"):
-            transcript = transcribe(uploaded_file, return_timestamps)
-            st.subheader("Transcription")
-            st.text_area("Transcription", transcript, height=200)
+            # Clean up temporary folder
+            for file_name in os.listdir(temp_folder):
+                file_path = os.path.join(temp_folder, file_name)
+                os.remove(file_path)
+            os.rmdir(temp_folder)
 
 
 if __name__ == "__main__":
@@ -82,3 +71,4 @@ if __name__ == "__main__":
 
 
 
+