File size: 2,282 Bytes
f23e3b4 fdd75f5 f23e3b4 fdd75f5 f23e3b4 fdd75f5 f23e3b4 fdd75f5 f23e3b4 fdd75f5 f23e3b4 fdd75f5 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 |
import os
os.environ["NUMBA_DISABLE_CACHE"] = "1"
import streamlit as st
import whisper
from gtts import gTTS
from moviepy.editor import VideoFileClip, AudioFileClip
from tempfile import NamedTemporaryFile
import torchaudio
st.set_page_config(page_title="AI Voiceover", layout="centered")
st.title("🎤 AI Voiceover App")
@st.cache_resource
def load_whisper_model():
return whisper.load_model("small")
whisper_model = load_whisper_model()
video_file = st.file_uploader("Upload a short video (MP4 preferred)", type=["mp4", "mov", "avi"])
if video_file:
with NamedTemporaryFile(delete=False, suffix=".mp4") as tmp_video:
tmp_video.write(video_file.read())
tmp_video_path = tmp_video.name
st.video(tmp_video_path)
video = VideoFileClip(tmp_video_path)
audio_path = tmp_video_path.replace(".mp4", ".wav")
video.audio.write_audiofile(audio_path)
st.info("Transcribing using Whisper...")
result = whisper_model.transcribe(audio_path)
st.subheader("📝 Detected Speech")
st.write(result["text"])
custom_text = st.text_area("Enter your voiceover text:", result["text"])
if st.button("Generate AI Voiceover"):
ai_voice_path = audio_path.replace(".wav", "_ai_voice.wav")
tts = gTTS(text=custom_text)
tts.save(ai_voice_path)
st.audio(ai_voice_path)
original_audio, sr = torchaudio.load(audio_path)
ai_audio, _ = torchaudio.load(ai_voice_path)
if ai_audio.shape[1] < original_audio.shape[1]:
diff = original_audio.shape[1] - ai_audio.shape[1]
ai_audio = torchaudio.functional.pad(ai_audio, (0, diff))
else:
ai_audio = ai_audio[:, :original_audio.shape[1]]
mixed_audio = (original_audio * 0.4) + (ai_audio * 0.6)
mixed_path = audio_path.replace(".wav", "_mixed.wav")
torchaudio.save(mixed_path, mixed_audio, sr)
final_video = video.set_audio(AudioFileClip(mixed_path))
final_path = tmp_video_path.replace(".mp4", "_final_streamlit.mp4")
final_video.write_videofile(final_path, codec="libx264", audio_codec="aac")
with open(final_path, "rb") as f:
st.download_button(label="📥 Download Final Video", data=f, file_name="final_ai_voiceover.mp4")
|