<enhance>: modified inference codes

1. Can set bbox_shift in configs/inference/test.yaml
2. Do not need to pip install whisper now

musetalk/utils/blending.py

@@ -52,7 +52,6 @@ def get_image(image,face,face_box,upper_boundary_ratio = 0.5,expand=1.2):
     blur_kernel_size = int(0.1 * ori_shape[0] // 2 * 2) + 1
     mask_array = cv2.GaussianBlur(np.array(modified_mask_image), (blur_kernel_size, blur_kernel_size), 0)
     mask_image = Image.fromarray(mask_array)
-    mask_image.save("./debug_mask.png")
     
     face_large.paste(face, (x-x_s, y-y_s, x1-x_s, y1-y_s))
     body.paste(face_large, crop_box[:2], mask_image)

musetalk/whisper/audio2feature.py

@@ -1,7 +1,5 @@
 import os
-#import whisper
-from whisper import load_model
-#import whisper.whispher as whiisper
+from .whisper import load_model
 import soundfile as sf
 import numpy as np
 import time
@@ -9,11 +7,12 @@ import sys
 sys.path.append("..")
 
 class Audio2Feature():
-    def __init__(self, whisper_model_type="tiny",model_path="./checkpoints/wisper_tiny.pt"):
+    def __init__(self, 
+                 whisper_model_type="tiny",
+                 model_path="./models/whisper/tiny.pt"):
         self.whisper_model_type = whisper_model_type
         self.model = load_model(model_path) #
 
-
     def get_sliced_feature(self,feature_array, vid_idx, audio_feat_length= [2,2],fps = 25):
         """
         Get sliced features based on a given index

musetalk/whisper/requirements.txt

@@ -1,6 +0,0 @@
-numpy
-torch
-tqdm
-more-itertools
-transformers>=4.19.0
-ffmpeg-python==0.2.0

musetalk/whisper/setup.py

@@ -1,24 +0,0 @@
-import os
-
-import pkg_resources
-from setuptools import setup, find_packages
-
-setup(
-    name="whisper",
-    py_modules=["whisper"],
-    version="1.0",
-    description="",
-    author="OpenAI",
-    packages=find_packages(exclude=["tests*"]),
-    install_requires=[
-        str(r)
-        for r in pkg_resources.parse_requirements(
-            open(os.path.join(os.path.dirname(__file__), "requirements.txt"))
-        )
-    ],
-    entry_points = {
-        'console_scripts': ['whisper=whisper.transcribe:cli'],
-    },
-    include_package_data=True,
-    extras_require={'dev': ['pytest']},
-)

musetalk/whisper/whisper.egg-info/PKG-INFO

@@ -1,5 +0,0 @@
-Metadata-Version: 2.1
-Name: whisper
-Version: 1.0
-Author: OpenAI
-Provides-Extra: dev

musetalk/whisper/whisper.egg-info/SOURCES.txt

@@ -1,18 +0,0 @@
-setup.py
-whisper/__init__.py
-whisper/__main__.py
-whisper/audio.py
-whisper/decoding.py
-whisper/model.py
-whisper/tokenizer.py
-whisper/transcribe.py
-whisper/utils.py
-whisper.egg-info/PKG-INFO
-whisper.egg-info/SOURCES.txt
-whisper.egg-info/dependency_links.txt
-whisper.egg-info/entry_points.txt
-whisper.egg-info/requires.txt
-whisper.egg-info/top_level.txt
-whisper/normalizers/__init__.py
-whisper/normalizers/basic.py
-whisper/normalizers/english.py

musetalk/whisper/whisper.egg-info/dependency_links.txt

@@ -1 +0,0 @@
-

musetalk/whisper/whisper.egg-info/entry_points.txt

@@ -1,2 +0,0 @@
-[console_scripts]
-whisper = whisper.transcribe:cli

musetalk/whisper/whisper.egg-info/requires.txt

@@ -1,9 +0,0 @@
-numpy
-torch
-tqdm
-more-itertools
-transformers>=4.19.0
-ffmpeg-python==0.2.0
-
-[dev]
-pytest

musetalk/whisper/whisper.egg-info/top_level.txt

@@ -1 +0,0 @@
-whisper