API reference / YouTube / Video transcript
Video transcript
Full transcript of a YouTube video with timed segments, word count, keywords, key points and sentiment.
5 credits per successful call
Errors are never charged
Request
GET
https://socialgraph.dev/api/v1/youtube/transcript
Send your API key as Authorization: Bearer YOUR_API_KEY.
Parameters go in the query string.
Parameters
| Name | Type | Description |
|---|---|---|
video_idrequired |
string | YouTube video ID, the part after v= in the URL, e.g. OYp5j2RihPE. |
languagesoptional |
string | Preferred transcript language code, e.g. en or hi. Defaults to en. If that language isn't available, another available transcript is returned; check language_code in the response. |
Code examples
Example response
{
"url": "https://www.youtube.com/watch?v=OYp5j2RihPE",
"keywords": [
"image",
"antigravity",
"agent",
"geometry",
"blender",
"gemini",
"argon",
"lighting",
"prompt",
"model"
],
"language": "English",
"video_id": "OYp5j2RihPE",
"sentiment": {
"label": "neutral",
"neutral": 0.9796,
"compound": 0.0059,
"negative": 0.0102,
"positive": 0.0103,
"sentence_counts": {
"neutral": 10,
"negative": 1,
"positive": 1
}
},
"conclusion": "Can an AI agent look at a 2D image that uses complex architectural geometry and build it inside Blender on its own? Next, I fed that image into my terminal connecting Antigravity directly to Blender via the MCP. An agent even automatically overlaid these comparison strips against the reference image to get the geometry closer, iteration after iteration.",
"key_points": [
"Can an AI agent look at a 2D image that uses complex architectural geometry and build it inside Blender on its own?",
"Next, I fed that image into my terminal connecting Antigravity directly to Blender via the MCP.",
"An agent even automatically overlaid these comparison strips against the reference image to get the geometry closer, iteration after iteration.",
"The model automatically dialed in volumetric fog, camera depth and audio-reactive sunlight And here's the final result: from a flat 2D prompt to a fully realized light-mapped, 3D cathedral walkthrough.",
"All driven by Gemini 4 Argon in Antigravity."
],
"transcript": {
"text": "Can an AI agent look at a 2D image that uses complex architectural geometry and build it inside Blender on its own? This is how I use Gemini 4 Argon and Antigravity to do so. I started in Flow, prompting for an empty catherdral with specific lighting and angles. I chose this particular image of the repeating arches and columns that creates this perspective effect. Next, I fed that image into my terminal connecting Antigravity directly to Blender via the MCP. I kicked off an agentic loop prompt. For 45 rounds, the model iteratively built the base mesh. An agent even automatically overlaid these comparison strips against the reference image to get the geometry closer, iteration after iteration. Once the structure was locked, I paused the CLI to trigger target targeted micro-loops for column fluting and floor textures. Then came lighting. The model automatically dialed in volumetric fog, camera depth and audio-reactive sunlight And here's the final result: from a flat 2D prompt to a fully realized light-mapped, 3D cathedral walkthrough. All driven by Gemini 4 Argon in Antigravity.",
"segments": [
{
"text": "Can an AI agent look at a 2D image",
"start": 0,
"duration": 2.96
},
{
"text": "that uses complex architectural geometry",
"start": 2.96,
"duration": 2.46
},
{
"text": " and build it inside Blender on its own?",
"start": 5.42,
"duration": 1.71
},
{
"text": "This is how I use Gemini 4 Argon\nand Antigravity to do so.",
"start": 7.13,
"duration": 3.54
},
{
"text": "I started in Flow, prompting for an empty catherdral",
"start": 10.68,
"duration": 3.17
},
{
"text": "with specific lighting and angles.",
"start": 13.85,
"duration": 2.67
},
{
"text": "I chose this particular image",
"start": 16.52,
"duration": 2.13
},
{
"text": "of the repeating arches and columns",
"start": 18.64,
"duration": 2
},
{
"text": "that creates this perspective effect.",
"start": 20.64,
"duration": 1.83
},
{
"text": "Next, I fed that image into my terminal",
"start": 22.48,
"duration": 2.67
},
{
"text": "connecting Antigravity directly to Blender via the MCP.",
"start": 25.15,
"duration": 4.09
},
{
"text": "I kicked off an agentic loop prompt.",
"start": 29.24,
"duration": 1.79
},
{
"text": "For 45 rounds, the model iteratively built the base mesh.",
"start": 31.03,
"duration": 3.71
},
{
"text": "An agent even automatically overlaid",
"start": 34.74,
"duration": 2.46
},
{
"text": "these comparison strips",
"start": 37.2,
"duration": 1.33
},
{
"text": "against the reference image",
"start": 38.54,
"duration": 1.58
},
{
"text": "to get the geometry closer,",
"start": 40.12,
"duration": 1.54
},
{
"text": "iteration after iteration.",
"start": 41.67,
"duration": 1.88
},
{
"text": "Once the structure was locked,",
"start": 43.54,
"duration": 1.42
},
{
"text": "I paused the CLI to trigger target targeted micro-loops",
"start": 44.96,
"duration": 2.5
},
{
"text": "for column fluting and floor textures.",
"start": 47.46,
"duration": 2.42
},
{
"text": "Then came lighting.",
"start": 49.88,
"duration": 1.46
},
{
"text": "The model automatically dialed in volumetric fog,",
"start": 51.34,
"duration": 3.21
},
{
"text": "camera depth and audio-reactive sunlight",
"start": 54.55,
"duration": 2.5
},
{
"text": "And here's the final result:",
"start": 57.35,
"duration": 1.5
},
{
"text": "from a flat 2D prompt",
"start": 58.85,
"duration": 1.71
},
{
"text": "to a fully realized",
"start": 60.56,
"duration": 1.29
},
{
"text": "light-mapped, 3D cathedral walkthrough.",
"start": 61.85,
"duration": 2.21
},
{
"text": "All driven by Gemini 4 Argon in Antigravity.",
"start": 64.06,
"duration": 3.59
}
]
},
"word_count": 179,
"is_generated": false,
"language_code": "en",
"segment_count": 29,
"duration_seconds": 67.65,
"requested_languages": [
"en"
]
}Errors
| Status | Meaning | Charged? |
|---|---|---|
| 401 | Missing, invalid or revoked API key | No |
| 402 | Not enough credits (insufficient_credits) | No |
| 404 | The account or post doesn't exist | No |
| 422 | Missing or invalid parameters | No |
| 502 / 503 | The data source is temporarily unavailable | No |
Try it live: Video transcript
GET
🔒 Preset demo input. Live demos use the Google for Developers channel, youtube.com/@GoogleDevelopers. Sign up free to use your own.
More YouTube endpoints
GET
Activities list
Read-only YouTube Data API v3 activities list endpoint.
/api/v1/youtube/activities
1 credit / call
GET
Channel sections list
Read-only YouTube Data API v3 channelSections list endpoint.
/api/v1/youtube/channelSections
1 credit / call
GET
Channels list
Read-only YouTube Data API v3 channels list endpoint.
/api/v1/youtube/channels
1 credit / call
GET
Comment threads list
Read-only YouTube Data API v3 commentThreads list endpoint.
/api/v1/youtube/commentThreads
1 credit / call
GET
Comments list
Read-only YouTube Data API v3 comments list endpoint.
/api/v1/youtube/comments
1 credit / call
GET
Internationalization languages list
Read-only YouTube Data API v3 i18nLanguages list endpoint.
/api/v1/youtube/i18nLanguages
1 credit / call