Upload 4 files

Dockerfile

@@ -0,0 +1,26 @@
+FROM python:3.10-slim
+
+ENV PIP_NO_CACHE_DIR=true \
+    PYTHONDONTWRITEBYTECODE=1 \
+    PYTHONUNBUFFERED=1 \
+    NUMBA_DISABLE_CACHE=1
+
+RUN apt-get update && apt-get install -y \
+    git ffmpeg libsndfile1 build-essential python3-dev libffi-dev wget curl \
+    && apt-get clean
+
+WORKDIR /app
+
+COPY app.py app.py
+COPY requirements.txt requirements.txt
+COPY README.md README.md
+
+RUN pip install --upgrade pip setuptools wheel
+RUN pip install numpy==1.24.3
+RUN pip install git+https://github.com/openai/whisper.git
+RUN pip install -r requirements.txt
+RUN pip uninstall -y librosa
+RUN pip install git+https://github.com/kamperh/librosa-no-numba.git
+RUN pip install git+https://github.com/coqui-ai/TTS.git
+
+CMD ["streamlit", "run", "app.py", "--server.port=7860", "--server.address=0.0.0.0"]

README.md

@@ -1,10 +1,13 @@
 ---
-title: Video0004
-emoji: 🏃
-colorFrom: purple
-colorTo: gray
+title: AI Voiceover App V2 (Docker)
+emoji: 🧠
+colorFrom: gray
+colorTo: blue
 sdk: docker
 pinned: false
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# 🧠 AI Voiceover V2 — Replace One Speaker Only
+
+Runs Whisper + Coqui TTS + audio mixing inside a Docker container with patched Librosa and Numba fixes for Hugging Face.
+

app.py

@@ -0,0 +1,68 @@
+import os
+os.environ["NUMBA_DISABLE_CACHE"] = "1"
+
+import streamlit as st
+import whisper
+from TTS.api import TTS
+from moviepy.editor import VideoFileClip, AudioFileClip, CompositeAudioClip
+from tempfile import NamedTemporaryFile
+import torchaudio
+
+st.set_page_config(page_title="AI Voiceover Generator V2", layout="centered")
+st.title("🎤 AI Voiceover V2: Replace One Speaker Only")
+
+@st.cache_resource
+def load_whisper_model():
+    return whisper.load_model("small")
+
+@st.cache_resource
+def load_tts_model():
+    return TTS(model_name="tts_models/en/ljspeech/tacotron2-DDC", progress_bar=False, gpu=False)
+
+whisper_model = load_whisper_model()
+tts = load_tts_model()
+
+video_file = st.file_uploader("Upload a short video clip (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 custom voiceover text to replace one speaker:", result["text"])
+
+    if st.button("Replace Only One Speaker's Voice"):
+        ai_voice_path = audio_path.replace(".wav", "_ai_voice.wav")
+        tts.tts_to_file(text=custom_text, file_path=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_v2.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 with Mixed Voiceover", data=f, file_name="final_ai_video_v2.mp4")

requirements.txt

@@ -0,0 +1,6 @@
+streamlit
+moviepy
+ffmpeg-python
+torchaudio
+torch
+numba