Skip to content
Python

fixed silent audio outputs for infinitalk/multitalk ...

DeepBeepMeepbaeda336bff1
diff --git a/wgp.py b/wgp.py
index 6a26750..6407757 100644
--- a/wgp.py
+++ b/wgp.py
@@ -5710,7 +5710,6 @@ def generate_video(
source_video_frames_count = 0 # number of frames to use in source video (processing starts source_video_overlap_frames_count frames before )
frames_already_processed = []
frames_already_processed_count = 0
- output_new_audio_filepath = None
overlapped_latents = None
context_scale = None
window_no = 0
@@ -6209,11 +6208,12 @@ def generate_video(
overlapped_latents = samples.get("latent_slice", None)
BGRA_frames = samples.get("BGRA_frames", None)
generated_audio = samples.get("audio", generated_audio)
- if generated_audio is not None and model_def.get("output_audio_is_input_audio", False) and output_new_audio_filepath is not None:
- generated_audio = None
- else:
- output_new_audio_filepath = None
- output_audio_sampling_rate = samples.get("audio_sampling_rate", audio_sampling_rate)
+ if generated_audio is not None:
+ if model_def.get("output_audio_is_input_audio", False) and output_new_audio_filepath is not None:
+ generated_audio = None
+ else:
+ output_new_audio_filepath = None
+ output_audio_sampling_rate = samples.get("audio_sampling_rate", audio_sampling_rate)
post_decode_pre_trim = samples.get("post_decode_pre_trim", 0)
samples = samples.get("x", None)
@@ -6369,7 +6369,8 @@ def generate_video(
new_audio_tracks = [output_new_audio_filepath]
else:
new_audio_tracks = control_audio_tracks
-
+ if generated_audio is not None: output_new_audio_filepath = None
+
combine_and_concatenate_video_with_audio_tracks(
video_path,
save_path_tmp,