nodes/havvk-ComfyUI_AIIA
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

v1.9.101: Robust Loop-Top Validation for Speech-Only Blink (Fixed silence-blinks after skips)

Browse Source
This commit is contained in:
Hawk Lee
2026-01-23 22:58:41 +08:00
parent db0816f813
commit 858f9dbb65
2 changed files with 46 additions and 47 deletions
Download Patch File Download Diff File Expand all files Collapse all files
libs/Ditto/core/atomic_components/motion_stitch.py
+45 -46
View File
@@ -187,6 +187,39 @@ def _set_eye_blink_idx(N, blink_n=15, open_n=-1, interval_min=60, interval_max=1
blink_hold_max = 1
while cur_i < max_i:
# [v1.9.101] Strict Speech-Only Validation
# Check if current cur_i is valid for a blink
if speech_only_blink and vad_timeline is not None:
# If current position is silence, find next segment
if vad_timeline[cur_i] <= 0.1:
scan_start = cur_i
while scan_start < len(vad_timeline) and vad_timeline[scan_start] <= 0.1:
scan_start += 1
if scan_start >= len(vad_timeline): break
# New segment found
scan_end = scan_start
while scan_end < len(vad_timeline) and vad_timeline[scan_end] > 0.1:
scan_end += 1
# Force blink into this segment
if (scan_end - scan_start) >= blink_n:
cur_i = random.randint(scan_start, scan_end - blink_n)
else:
cur_i = scan_end
continue # Too short, try next
else:
# Already in speech, check if it fits before segment ends
scan_end = cur_i
while scan_end < len(vad_timeline) and vad_timeline[scan_end] > 0.1:
scan_end += 1
if cur_i + blink_n > scan_end:
# Not enough room here, jump to end and find next segment
cur_i = scan_end
continue
# First Blink
hold_frames = random.randint(blink_hold_min, blink_hold_max)
@@ -236,59 +269,25 @@ def _set_eye_blink_idx(N, blink_n=15, open_n=-1, interval_min=60, interval_max=1
idx[cur_i : cur_i + full_len_2] = full_blink_seq_2
cur_i += full_len_2
# [v1.9.92] Speech-Only Blink: Only insert blinks during speech.
# This masks mesh coupling artifacts naturally.
# Calculate Next Event
if open_ns:
cur_n = open_ns[cur_n_i % len(open_ns)]
cur_n_i += 1
else:
if vad_timeline is not None:
if speech_only_blink:
# Speech-Only Mode: Scan ahead to find next speech segment
scan_start = cur_i
while scan_start < len(vad_timeline) and vad_timeline[scan_start] <= 0.1:
scan_start += 1
if scan_start >= len(vad_timeline):
print(f" [Blink Loop] No more speech found after frame {cur_i}. Stopping.")
break
# Find end of this segment
scan_end = scan_start
while scan_end < len(vad_timeline) and vad_timeline[scan_end] > 0.1:
scan_end += 1
# Next blink: interval_min to interval_max
jump = random.randint(OPEN_MIN, OPEN_MAX)
target_i = scan_start + jump
if target_i + blink_n <= scan_end:
cur_i = target_i
print(f" [Blink Loop] Speech-Only: Next speech at {scan_start}-{scan_end}, jumping +{jump} to {cur_i}")
else:
if (scan_end - scan_start) >= blink_n:
cur_i = random.randint(scan_start, scan_end - blink_n)
print(f" [Blink Loop] Speech-Only: Target {target_i} overshot segment end {scan_end}. Pulling back to {cur_i}")
else:
print(f" [Blink Loop] Speech-Only: Segment at {scan_start}-{scan_end} too short (<{blink_n}). Skipping.")
cur_i = scan_end
cur_n = 0
continue # Try next segment
cur_n = 0
if vad_timeline is not None and not speech_only_blink:
# Natural Mode (Not speech-only)
is_speaking = vad_timeline[cur_i] > 0.1 if cur_i < len(vad_timeline) else False
if is_speaking:
cur_n = random.randint(OPEN_MIN, OPEN_MAX)
print(f" [Blink Loop] Natural: Speaking at {cur_i}, next in {cur_n}")
else:
# Natural Mode: Scale user intervals based on speech
is_speaking = vad_timeline[cur_i] > 0.1 if cur_i < len(vad_timeline) else False
if is_speaking:
cur_n = random.randint(OPEN_MIN, OPEN_MAX)
print(f" [Blink Loop] Natural: Speaking at {cur_i}, next in {cur_n}")
else:
# Silence: Relaxed Interval (1.5x longer)
cur_n = random.randint(int(OPEN_MIN * 1.5), int(OPEN_MAX * 1.5))
print(f" [Blink Loop] Natural: Silence at {cur_i}, next in {cur_n} (1.5x Relaxed)")
cur_n = random.randint(int(OPEN_MIN * 1.5), int(OPEN_MAX * 1.5))
print(f" [Blink Loop] Natural: Silence at {cur_i}, next in {cur_n} (1.5x Relaxed)")
else:
# Standard or Speech-Only prep (Speech-Only will re-validate at loop top)
cur_n = random.randint(OPEN_MIN, OPEN_MAX)
print(f" [Blink Loop] Standard (No VAD): Next in {cur_n}")
if speech_only_blink:
print(f" [Blink Loop] Speech-Only Pre-Interval: +{cur_n}")
cur_i += cur_n
pyproject.toml
+1 -1
View File
@@ -1,7 +1,7 @@
[project]
name = "aiia"
description = "The Ultimate AI Audio/Video toolkit for ComfyUI. Features an enhanced Ditto (with optimizations that outperform official demos and other SOTA talking head models in lip-sync accuracy and natural motion), EchoMimic V3 & FLOAT, VibeVoice & CosyVoice 3.0 (Zero-Shot Voice Cloning), Multi-Role Podcast Generation, and a powerful Media Browser."
version = "1.9.100"
version = "1.9.101"
license = {file = "LICENSE"}
readme = "README.md"
authors = [