Testing application without colab

[iboyles]

I trained the model locally finally on a 5 minute video that had eye contact and the live asr gui seems to have too much latency for me to test accuracy. How can I display audio next to the video on my local machine like in the colab the code chunk that plays them together on colab does not transfer to python directly :

@title Display Video

import os import glob from IPython.display import HTML from base64 import b64encode

def get_latest_file(path): dir_list = glob.glob(path) dir_list.sort(key=lambda x: os.path.getmtime(x)) return dir_list[-1]

Video = get_latest_file(os.path.join('trial_ian', 'results', '*.mp4')) Video_aud = Video.replace('.mp4', 'aud.mp4')

concat audio

! ffmpeg -y -i {Video} -i data/{Aud} -c:v copy -c:a aac {Video_aud}

display

def show_video(video_path, video_width=450):

video_file = open(video_path, "r+b").read() video_url = f"data:video/mp4;base64,{b64encode(video_file).decode()}"

return HTML(f"""""")

show_video(Video_aud)

[iboyles]

from moviepy.editor import VideoFileClip, AudioFileClip

def merge_audio_and_video(video_path, audio_path, output_path):
    video_clip = VideoFileClip(video_path)
    audio_clip = AudioFileClip(audio_path)

    # Make sure audio duration matches the video duration
    if video_clip.duration < audio_clip.duration:
        audio_clip = audio_clip.subclip(0, video_clip.duration)

    # Set the audio of the video clip to the merged audio
    video_clip = video_clip.set_audio(audio_clip)

    # Write the merged video with the new audio to the output path
    video_clip.write_videofile(output_path, codec="libx264")

    # Close the clips to free up resources
    video_clip.close()
    audio_clip.close()

if __name__ == "__main__":
    video_path = "trial_ian2_torso/results/ngp_ep0030.mp4"
    audio_path = "data/audio5.wav"
    output_path = "trial_ian2_torso/results/ian_aud.mp4"

    merge_audio_and_video(video_path, audio_path, output_path)

jk chat gpt wrote this script for me it tests it in python without ffmpeg may even be faster too.

[gautham-balraj]

Iam facing errors when running the model in collab,  can u help rocessing ./freqencoder Preparing metadata (setup.py) ... done Building wheels for collected packages: freqencoder error: subprocess-exited-with-error

× python setup.py bdist_wheel did not run successfully. │ exit code: 1 ╰─> See above for output.

note: This error originates from a subprocess, and is likely not a problem with pip. Building wheel for freqencoder (setup.py) ... error ERROR: Failed building wheel for freqencoder Running setup.py clean for freqencoder Failed to build freqencoder ERROR: Could not build wheels for freqencoder, which is required to install pyproject.toml-based projects Processing ./shencoder Preparing metadata (setup.py) ... done Building wheels for collected packages: shencoder