Spaces:
Running
on
Zero
Running
on
Zero
animikhaich
commited on
Merge branch 'web-app-dev' of github.com:animikhaich/VidTune into web-app-dev
Browse files- engine/audio_generator.py +8 -2
- engine/video_descriptor.py +4 -2
engine/audio_generator.py
CHANGED
@@ -72,8 +72,14 @@ class GenerateAudio:
|
|
72 |
prompts = self.prompts_sanity_check(prompts)
|
73 |
|
74 |
try:
|
75 |
-
|
76 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
77 |
self.result = result.cpu().numpy().T
|
78 |
self.result = self.result.transpose((2, 0, 1))
|
79 |
self.sampling_rate = self.model.sample_rate
|
|
|
72 |
prompts = self.prompts_sanity_check(prompts)
|
73 |
|
74 |
try:
|
75 |
+
if duration <= 30:
|
76 |
+
self.model.set_generation_params(duration=duration)
|
77 |
+
result = self.model.generate(prompts, progress=False)
|
78 |
+
elif duration > 30:
|
79 |
+
self.model.set_generation_params(duration=30)
|
80 |
+
result = self.model.generate(prompts, progress=False)
|
81 |
+
self.model.set_generation_params(duration=duration)
|
82 |
+
result = self.model.generate_with_chroma(prompts, result, melody_sample_rate=self.sampling_rate, progress=False)
|
83 |
self.result = result.cpu().numpy().T
|
84 |
self.result = self.result.transpose((2, 0, 1))
|
85 |
self.sampling_rate = self.model.sample_rate
|
engine/video_descriptor.py
CHANGED
@@ -47,7 +47,7 @@ class DescribeVideo:
|
|
47 |
|
48 |
logging.info(f"Initialized DescribeVideo with model: {self.model}")
|
49 |
|
50 |
-
def describe_video(self, video_path):
|
51 |
video_file = genai.upload_file(video_path)
|
52 |
logging.info(f"Uploaded video: {video_path}")
|
53 |
|
@@ -58,9 +58,11 @@ class DescribeVideo:
|
|
58 |
if video_file.state.name == "FAILED":
|
59 |
logging.error(f"Failed to upload video: {video_file.state.name}")
|
60 |
raise ValueError(f"Failed to upload video: {video_file.state.name}")
|
|
|
|
|
61 |
|
62 |
response = self.mllm_model.generate_content(
|
63 |
-
[video_file, "Explain what is happening in this video"],
|
64 |
request_options={"timeout": 600},
|
65 |
safety_settings=self.safety_settings,
|
66 |
)
|
|
|
47 |
|
48 |
logging.info(f"Initialized DescribeVideo with model: {self.model}")
|
49 |
|
50 |
+
def describe_video(self, video_path, genre, bpm, user_keywords):
|
51 |
video_file = genai.upload_file(video_path)
|
52 |
logging.info(f"Uploaded video: {video_path}")
|
53 |
|
|
|
58 |
if video_file.state.name == "FAILED":
|
59 |
logging.error(f"Failed to upload video: {video_file.state.name}")
|
60 |
raise ValueError(f"Failed to upload video: {video_file.state.name}")
|
61 |
+
|
62 |
+
additional_keywords = ", ".join([genre, user_keywords, bpm]) + "bpm"
|
63 |
|
64 |
response = self.mllm_model.generate_content(
|
65 |
+
[video_file, f"Explain what is happening in this video. The following keywords are provided by the user for generating the music prompt: {additional_keywords}"],
|
66 |
request_options={"timeout": 600},
|
67 |
safety_settings=self.safety_settings,
|
68 |
)
|