工作流 /
视频工具 /
VEED Dubbing：视频翻译配音
VEED Dubbing：视频翻译配音
VEED Dubbing: Video Translation
用 VEED Dubbing 把视频里的人声翻译成另一种语言：每个说话人的音色会被克隆，口型重新对齐译制后的台词。standard 档更便宜，pro 档增加地区口音与立体声。适合画面里有清晰人脸、音轨里有人声的素材。
视频工具
4 个节点 1 个输入 41 次使用 ComfyUI ≥ 0.43.1 ComfyUI 7 KB
节点构成
LoadVideo
× 1
SaveVideo
× 1
MarkdownNote
× 1
VeedDubbingNode
× 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": "47bc3771-0160-403a-adf1-d5bc4a986b4b",
3 "revision": 0,
4 "last_node_id": 5,
5 "last_link_id": 4,
6 "nodes": [
7 {
8 "id": 1,
9 "type": "LoadVideo",
10 "pos": [
11 100,
12 127.51953831643175
13 ],
14 "size": [
15 349.8534373297398,
16 461.69348994880534
17 ],
18 "flags": [],
19 "order": 0,
20 "mode": 0,
21 "inputs": [],
22 "outputs": [
23 {
24 "name": "VIDEO",
25 "type": "VIDEO",
26 "links": [
27 3
28 ]
29 }
30 ],
31 "properties": {
32 "Node name for S&R": "LoadVideo"
33 },
34 "widgets_values": [
35 "api_veed_dubbing_input_video.mp4",
36 "image"
37 ],
38 "widgets_values_named": {
39 "file": "api_veed_dubbing_input_video.mp4",
40 "upload": "image"
41 }
42 },
43 {
44 "id": 3,
45 "type": "SaveVideo",
46 "pos": [
47 825.1307810797398,
48 127.51953831643175
49 ],
50 "size": [
51 364.5434347231959,
52 461.69348994880534
53 ],
54 "flags": [],
55 "order": 3,
56 "mode": 0,
57 "inputs": [
58 {
59 "name": "video",
60 "type": "VIDEO",
61 "link": 4
62 }
63 ],
64 "outputs": [
65 {
66 "name": "video",
67 "type": "VIDEO",
68 "links": null
69 }
70 ],
71 "properties": [],
72 "widgets_values": [
73 "video/veed_dubbing",
74 "mp4",
75 "auto",
76 "h264",
77 "auto"
78 ],
79 "widgets_values_named": {
80 "filename_prefix": "video/veed_dubbing",
81 "format": "mp4",
82 "format.codec": "auto",
83 "codec": "h264",
84 "codec.encoding": "auto"
85 }
86 },
87 {
88 "id": 4,
89 "type": "MarkdownNote",
90 "pos": [
91 -249.8374813000641,
92 128.939015511904
93 ],
94 "size": [
95 320.07232923408924,
96 454.6558221364456
97 ],
98 "flags": [],
99 "order": 1,
100 "mode": 0,
101 "inputs": [],
102 "outputs": [],
103 "properties": [],
104 "widgets_values": [
105 "# VEED Dubbing\n\n**Translate the speech in a video into another language.** Every speaker's voice is cloned and their lips are re-synced\nto the translation, so the result looks spoken in the target language rather than captioned. Partner (API) node: it runs\non Comfy's servers, needs an account, and nothing is rendered locally.\n\n## Input Asset\n\n- [api_veed_dubbing_input_video.mp4](https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/input/api_veed_dubbing_input_video.mp4)\n\n## Requirements\n\n- **The audio must contain voice that can be converted.** Narration, dialogue or an interview all work; music, sound\n effects or silence give the model nothing to translate.\n- **There must be a face in the video.** Dubbing works on what it can see - a single speaker facing the camera is the\n reliable case. Backs of heads, hands, product shots and crowd scenes do not dub.\n\n## How to use it\n\n- `model`: `standard` is the cheaper tier with mono audio (Spanish is Latin American, Portuguese is Brazilian);\n `pro` adds regional accents and stereo audio.\n- `target_language`: 85 languages on standard, 101 options on pro (regional variants such as `English (UK)` or\n `Chinese (Taiwan)`).\n- `source_language`: leave it on `auto` unless the model mishears the original; naming the spoken language improves\n accuracy on short clips and strong accents.\n- `seed` only decides whether the node re-runs - the result is non-deterministic either way.\n\n## Limits\n\n- Up to 599 seconds of video. The output is as long as its audio track and plays at 25 fps, so a 24 fps source becomes\n 25 fps.\n- One speaker is the reliable case. Multi-speaker footage is supported but voice assignment can drift.\n- No subtitle file is produced: the node returns a dubbed video only.\n"
106 ],
107 "widgets_values_named": {
108 "text": "# VEED Dubbing\n\n**Translate the speech in a video into another language.** Every speaker's voice is cloned and their lips are re-synced\nto the translation, so the result looks spoken in the target language rather than captioned. Partner (API) node: it runs\non Comfy's servers, needs an account, and nothing is rendered locally.\n\n## Input Asset\n\n- [api_veed_dubbing_input_video.mp4](https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/input/api_veed_dubbing_input_video.mp4)\n\n## Requirements\n\n- **The audio must contain voice that can be converted.** Narration, dialogue or an interview all work; music, sound\n effects or silence give the model nothing to translate.\n- **There must be a face in the video.** Dubbing works on what it can see - a single speaker facing the camera is the\n reliable case. Backs of heads, hands, product shots and crowd scenes do not dub.\n\n## How to use it\n\n- `model`: `standard` is the cheaper tier with mono audio (Spanish is Latin American, Portuguese is Brazilian);\n `pro` adds regional accents and stereo audio.\n- `target_language`: 85 languages on standard, 101 options on pro (regional variants such as `English (UK)` or\n `Chinese (Taiwan)`).\n- `source_language`: leave it on `auto` unless the model mishears the original; naming the spoken language improves\n accuracy on short clips and strong accents.\n- `seed` only decides whether the node re-runs - the result is non-deterministic either way.\n\n## Limits\n\n- Up to 599 seconds of video. The output is as long as its audio track and plays at 25 fps, so a 24 fps source becomes\n 25 fps.\n- One speaker is the reliable case. Multi-speaker footage is supported but voice assignment can drift.\n- No subtitle file is produced: the node returns a dubbed video only.\n"
109 },
110 "color": "#222",
111 "bgcolor": "#000"
112 },
113 {
114 "id": 5,
115 "type": "VeedDubbingNode",
116 "pos": [
117 491.20080124499304,
118 127.9509725586428
119 ],
120 "size": [
121 270,
122 154
123 ],
124 "flags": [],
125 "order": 2,
126 "mode": 0,
127 "inputs": [
128 {
129 "name": "model.video",
130 "type": "VIDEO",
131 "link": 3
132 }
133 ],
134 "outputs": [
135 {
136 "name": "VIDEO",
137 "type": "VIDEO",
138 "links": [
139 4
140 ]
141 }
142 ],
143 "properties": {
144 "Node name for S&R": "VeedDubbingNode"
145 },
146 "widgets_values": [
147 "standard",
148 "Japanese",
149 "auto",
150 42,
151 "randomize"
152 ],
153 "widgets_values_named": {
154 "model": "standard",
155 "model.target_language": "Japanese",
156 "model.source_language": "auto",
157 "model.seed": 42,
158 "control_after_generate": "randomize"
159 },
160 "color": "#432",
161 "bgcolor": "#653"
162 }
163 ],
164 "links": [
165 [
166 3,
167 1,
168 0,
169 5,
170 0,
171 "VIDEO"
172 ],
173 [
174 4,
175 5,
176 0,
177 3,
178 0,
179 "VIDEO"
180 ]
181 ],
182 "groups": [],
183 "config": [],
184 "extra": {
185 "ds": {
186 "scale": 1.012559034060532,
187 "offset": [
188 442.37115466460864,
189 263.023910539666
190 ]
191 },
192 "frontendVersion": "1.57.4"
193 },
194 "version": 0.4
195}