ComfyPilotComfyPilot
工作流 / 视频工具 / VEED Clean Audio：视频语音降噪
VEED Clean Audio：视频语音降噪

VEED Clean Audio：视频语音降噪

VEED Clean Audio: Video Speech Cleanup

用 VEED Clean Audio 清理视频里的语音：底噪、音乐与音效会一起从音轨中被移除，句间停顿被彻底清理，视频带着 48 kHz 单声道语音返回，并可按目标响度归一化，画面保持原样。单段视频最长 1799 秒。

视频工具 4 个节点 1 个输入 ComfyUI ≥ 0.43.0 ComfyUI 5 KB

下载工作流 JSON 查看官方来源

下载后把文件拖进 ComfyUI 窗口即可打开。

节点构成

VeedCleanVideoAudioNode × 1
LoadVideo × 1
SaveVideo × 1
MarkdownNote × 1

官方标注的模型

VEED

工作流 JSON

和上面「下载工作流 JSON」是同一份文件。

1{
2 "id": "c1aadf6c-4d94-4377-8621-8022846cea76",
3 "revision": 0,
4 "last_node_id": 4,
5 "last_link_id": 2,
6 "nodes": [
7 {
8 "id": 1,
9 "type": "VeedCleanVideoAudioNode",
10 "pos": [
11 2050,
12 730
13 ],
14 "size": [
15 330,
16 178
17 ],
18 "flags": [],
19 "order": 2,
20 "mode": 0,
21 "inputs": [
22 {
23 "name": "video",
24 "type": "VIDEO",
25 "link": 1
26 }
27 ],
28 "outputs": [
29 {
30 "name": "VIDEO",
31 "type": "VIDEO",
32 "links": [
33 2
34 ]
35 }
36 ],
37 "properties": {
38 "Node name for S&R": "VeedCleanVideoAudioNode"
39 },
40 "widgets_values": [
41 0.874,
42 "normalize",
43 -19
44 ],
45 "widgets_values_named": {
46 "strength": 0.874,
47 "loudness": "normalize",
48 "loudness.target_lufs": -19
49 },
50 "color": "#432",
51 "bgcolor": "#653"
52 },
53 {
54 "id": 2,
55 "type": "LoadVideo",
56 "pos": [
57 1620,
58 730
59 ],
60 "size": [
61 370,
62 460
63 ],
64 "flags": [],
65 "order": 0,
66 "mode": 0,
67 "inputs": [],
68 "outputs": [
69 {
70 "name": "VIDEO",
71 "type": "VIDEO",
72 "links": [
73 1
74 ]
75 }
76 ],
77 "properties": {
78 "Node name for S&R": "LoadVideo"
79 },
80 "widgets_values": [
81 "veed_clean_audio_input_h3_omniref_person_dialogue_plus_noise.mp4",
82 "image"
83 ],
84 "widgets_values_named": {
85 "file": "veed_clean_audio_input_h3_omniref_person_dialogue_plus_noise.mp4",
86 "upload": "image"
87 }
88 },
89 {
90 "id": 3,
91 "type": "SaveVideo",
92 "pos": [
93 2440,
94 730
95 ],
96 "size": [
97 520,
98 160
99 ],
100 "flags": [],
101 "order": 3,
102 "mode": 0,
103 "inputs": [
104 {
105 "name": "video",
106 "type": "VIDEO",
107 "link": 2
108 }
109 ],
110 "outputs": [
111 {
112 "name": "video",
113 "type": "VIDEO",
114 "links": null
115 }
116 ],
117 "properties": [],
118 "widgets_values": [
119 "video/VEED_Clean_Audio_Video",
120 "auto",
121 "auto",
122 "auto"
123 ],
124 "widgets_values_named": {
125 "filename_prefix": "video/VEED_Clean_Audio_Video",
126 "format": "auto",
127 "format.codec": "auto",
128 "codec": "auto"
129 }
130 },
131 {
132 "id": 4,
133 "type": "MarkdownNote",
134 "pos": [
135 1040,
136 730
137 ],
138 "size": [
139 520,
140 620
141 ],
142 "flags": [],
143 "order": 1,
144 "mode": 0,
145 "inputs": [],
146 "outputs": [],
147 "properties": [],
148 "widgets_values": [
149 "## VEED Clean Audio (in Video): Speech Cleanup — Workflow Guide\n\n## Input Asset\n- [veed_clean_audio_input_h3_omniref_person_dialogue_plus_noise.mp4](https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/input/veed_clean_audio_input_h3_omniref_person_dialogue_plus_noise.mp4)\n\n## How to Use\n1. Load a clip with an audio track.\n2. Keep strength at 0.874: how much noise is removed.\n3. Choose normalize (target -19 LUFS) or keep input level.\n\n## Limits\n- Up to 1799 seconds (just under 30 minutes) per run.\n- Audio returns 48 kHz mono; stereo is folded down.\n- A clip with no audio track stops with an error.\n- Peaks are capped at -1.1 dBTP, so loud targets settle near -15 LUFS.\n"
150 ],
151 "widgets_values_named": {
152 "text": "## VEED Clean Audio (in Video): Speech Cleanup — Workflow Guide\n\n## Input Asset\n- [veed_clean_audio_input_h3_omniref_person_dialogue_plus_noise.mp4](https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/input/veed_clean_audio_input_h3_omniref_person_dialogue_plus_noise.mp4)\n\n## How to Use\n1. Load a clip with an audio track.\n2. Keep strength at 0.874: how much noise is removed.\n3. Choose normalize (target -19 LUFS) or keep input level.\n\n## Limits\n- Up to 1799 seconds (just under 30 minutes) per run.\n- Audio returns 48 kHz mono; stereo is folded down.\n- A clip with no audio track stops with an error.\n- Peaks are capped at -1.1 dBTP, so loud targets settle near -15 LUFS.\n"
153 },
154 "color": "#222",
155 "bgcolor": "#000",
156 "title": "Note: Workflow Guide"
157 }
158 ],
159 "links": [
160 [
161 1,
162 2,
163 0,
164 1,
165 0,
166 "VIDEO"
167 ],
168 [
169 2,
170 1,
171 0,
172 3,
173 0,
174 "VIDEO"
175 ]
176 ],
177 "groups": [],
178 "config": [],
179 "extra": {
180 "ds": {
181 "scale": 0.7750026059432124,
182 "offset": [
183 -846.2585280525677,
184 -523.5299707583583
185 ]
186 },
187 "frontendVersion": "1.58.0"
188 },
189 "version": 0.4
190}