Open-source AI podcast clipper. Generate vertical clips with face tracking and burned-in captions. CLI, MCP server, web app.
Python ai-toolsauto-captionscaptionscliffmpegmcpmcp-serveropusclip-alternativepodcast-clipspodcastsshortform-videovertical-videowhisper
fix: face analysis crash when transcript has no speaker labels (#11)
Nika Siradzee5a00fdae52d
diff --git a/backend/services/face_analysis.py b/backend/services/face_analysis.pyindex 8eb6e0b..17fe762 100644--- a/backend/services/face_analysis.py+++ b/backend/services/face_analysis.py@@ -305,7 +305,7 @@ def analyze_faces(speaker_mappings[top_2[1]] = 1 - speaker_mappings[top_2[0]]# Extra speakers → dominant speaker's cluster- dominant_idx = speaker_mappings.get(speakers_by_talk[0], 0)+ dominant_idx = speaker_mappings.get(speakers_by_talk[0], 0) if speakers_by_talk else 0for sp in speakers_by_talk[2:]:speaker_mappings[sp] = dominant_idxelif len(clusters_list) >= 2 and len(speakers) >= 2:diff --git a/backend/services/transcription.py b/backend/services/transcription.pyindex 3dde47d..69f7493 100644--- a/backend/services/transcription.py+++ b/backend/services/transcription.py@@ -9,6 +9,7 @@ Produces word-level timestamps with speaker labels by:import osimport subprocess+import sysimport tempfilefrom typing import Optional, Callable