> ## Documentation Index > Fetch the complete documentation index at: https://gcore.com/docs/llms.txt > Use this file to discover all available pages before exploring further. # Create AI-task > Creating an AI task. This method allows you to create an AI task for VOD video processing: - ASR: Transcribe video - ASR: Translate subtitles - CM: Sports detection - CM: Not Safe For Work (NSFW) content detection - CM: Soft nudity detection - CM: Hard nudity detection - CM: Objects recognition (soon) ![Auto generated subtitles example](https://demo-files.gvideo.io/apidocs/captions.gif) How to use: - Create an AI task, specify algoritm to use - Get `task_id` - Check a result using ```.../ai/tasks/{task_id}``` method For more detailed information, see the description of each method separately. **AI Automatic Speech Recognition (ASR)** AI is instrumental in automatic video processing for subtitles creation by using Automatic Speech Recognition (ASR) technology to transcribe spoken words into text, which can then be translated into multiple languages for broader accessibility. Categories: - ```transcription``` – to create subtitles/captions from audio in the original language. - ```translation``` – to transate subtitles/captions from the original language to 99+ other languages. AI subtitle transcription and translation tools are highly efficient, processing large volumes of audio-visual content quickly and providing accurate transcriptions and translations with minimal human intervention. Additionally, AI-driven solutions can significantly reduce costs and turnaround times compared to traditional methods, making them an invaluable resource for content creators and broadcasters aiming to reach global audiences. Example response with positive result: ``` { "status": "SUCCESS", "result": { "subtitles": [ { "start_time": "00:00:00.031", "end_time": "00:00:03.831", "text": "Come on team, ..." }, ... ] "vttContent": "WEBVTT\n\n1\n00:00:00.031 --> 00:00:03.831\nCome on team, ...", "concatenated_text": "Come on team, ...", "languages": [ "eng" ], "speech_detected": true } }, ... } ``` **AI Content Moderation (CM)** The AI Content Moderation API offers a powerful solution for analyzing video content to detect various categories of inappropriate material. Leveraging state-of-the-art AI models, this API ensures real-time analysis and flagging of sensitive or restricted content types, making it an essential tool for platforms requiring stringent content moderation. Categories: - ```nsfw```: Quick algorithm to detect pornographic material, ensuring content is "not-safe-for-work" or normal. - ```hard_nudity```: Detailed analisys of video which detects explicit nudity involving genitalia. - ```soft_nudity```: Detailed video analysis that reveals both explicit and partial nudity, including the presence of male and female faces and other uncovered body parts. - ```sport```: Recognizes various sporting activities. The AI Content Moderation API is an invaluable tool for managing and controlling the type of content being shared or streamed on your platform. By implementing this API, you can ensure compliance with community guidelines and legal requirements, as well as provide a safer environment for your users. Important notes: - It's allowed to analyse still images too (where applicable). Format of image: JPEG, PNG. In that case one image is the same as video of 1 second duration. - Not all frames in the video are used for analysis, but only key frames (Iframe). For example, if a key frame in a video is set every ±2 seconds, then detection will only occur at these timestamps. If an object appears and disappears between these time stamps, it will not be detected. We are working on a version to analyze more frames, please contact your manager or our support team to enable this method. Example response with positive result: ``` { "status": "SUCCESS", "result": { "nsfw_detected": true, "detection_results": [ "nsfw" ], "frames": [ { "label": "nsfw", "confidence": 1.0, "frame_number": 24 },... ] } } ``` **Additional information** Billing takes into account the duration of the analyzed video. Or the duration until the stop tag(where applicable), if the condition was triggered during the analysis. The heart of content moderation is AI, with additional services. They run on our own infrastructure, so the files/data are not transferred anywhere to external services. After processing, original files are also deleted from local storage of AI. Read more detailed information about our solution, and architecture, and benefits in the knowledge base and blog. ## OpenAPI ````yaml /api-reference/services_documented/streaming_api.yaml post /streaming/ai/tasks openapi: 3.1.0 info: title: Gcore OpenAPI – Streaming API description: >- This OpenAPI is an aggregated OpenAPI specification that unifies all Gcore products into a single file. It covers Cloud, CDN, DNS, WAAP, DDoS Protection, Object Storage, Streaming, and FastEdge services. version: '2026-05-14T07:00:22.640261+00:00' servers: - url: https://api.gcore.com security: - APIKey: [] tags: - name: AI x-displayName: AI - name: Broadcasts x-displayName: Broadcasts - name: Directories x-displayName: Directories - name: Overlays x-displayName: Overlays - name: Players x-displayName: Players - name: Playlists x-displayName: Playlists - name: QualitySets x-displayName: QualitySets - name: Restreams x-displayName: Restreams - name: Streaming Statistics x-displayName: Statistics - name: Streams x-displayName: Streams - name: Subtitles x-displayName: Subtitles - name: Videos x-displayName: Videos paths: /streaming/ai/tasks: post: tags: - AI summary: Create AI-task description: >- Creating an AI task. This method allows you to create an AI task for VOD video processing: - ASR: Transcribe video - ASR: Translate subtitles - CM: Sports detection - CM: Not Safe For Work (NSFW) content detection - CM: Soft nudity detection - CM: Hard nudity detection - CM: Objects recognition (soon) ![Auto generated subtitles example](https://demo-files.gvideo.io/apidocs/captions.gif) How to use: - Create an AI task, specify algoritm to use - Get `task_id` - Check a result using ```.../ai/tasks/{task_id}``` method For more detailed information, see the description of each method separately. **AI Automatic Speech Recognition (ASR)** AI is instrumental in automatic video processing for subtitles creation by using Automatic Speech Recognition (ASR) technology to transcribe spoken words into text, which can then be translated into multiple languages for broader accessibility. Categories: - ```transcription``` – to create subtitles/captions from audio in the original language. - ```translation``` – to transate subtitles/captions from the original language to 99+ other languages. AI subtitle transcription and translation tools are highly efficient, processing large volumes of audio-visual content quickly and providing accurate transcriptions and translations with minimal human intervention. Additionally, AI-driven solutions can significantly reduce costs and turnaround times compared to traditional methods, making them an invaluable resource for content creators and broadcasters aiming to reach global audiences. Example response with positive result: ``` { "status": "SUCCESS", "result": { "subtitles": [ { "start_time": "00:00:00.031", "end_time": "00:00:03.831", "text": "Come on team, ..." }, ... ] "vttContent": "WEBVTT\n\n1\n00:00:00.031 --> 00:00:03.831\nCome on team, ...", "concatenated_text": "Come on team, ...", "languages": [ "eng" ], "speech_detected": true } }, ... } ``` **AI Content Moderation (CM)** The AI Content Moderation API offers a powerful solution for analyzing video content to detect various categories of inappropriate material. Leveraging state-of-the-art AI models, this API ensures real-time analysis and flagging of sensitive or restricted content types, making it an essential tool for platforms requiring stringent content moderation. Categories: - ```nsfw```: Quick algorithm to detect pornographic material, ensuring content is "not-safe-for-work" or normal. - ```hard_nudity```: Detailed analisys of video which detects explicit nudity involving genitalia. - ```soft_nudity```: Detailed video analysis that reveals both explicit and partial nudity, including the presence of male and female faces and other uncovered body parts. - ```sport```: Recognizes various sporting activities. The AI Content Moderation API is an invaluable tool for managing and controlling the type of content being shared or streamed on your platform. By implementing this API, you can ensure compliance with community guidelines and legal requirements, as well as provide a safer environment for your users. Important notes: - It's allowed to analyse still images too (where applicable). Format of image: JPEG, PNG. In that case one image is the same as video of 1 second duration. - Not all frames in the video are used for analysis, but only key frames (Iframe). For example, if a key frame in a video is set every ±2 seconds, then detection will only occur at these timestamps. If an object appears and disappears between these time stamps, it will not be detected. We are working on a version to analyze more frames, please contact your manager or our support team to enable this method. Example response with positive result: ``` { "status": "SUCCESS", "result": { "nsfw_detected": true, "detection_results": [ "nsfw" ], "frames": [ { "label": "nsfw", "confidence": 1.0, "frame_number": 24 },... ] } } ``` **Additional information** Billing takes into account the duration of the analyzed video. Or the duration until the stop tag(where applicable), if the condition was triggered during the analysis. The heart of content moderation is AI, with additional services. They run on our own infrastructure, so the files/data are not transferred anywhere to external services. After processing, original files are also deleted from local storage of AI. Read more detailed information about our solution, and architecture, and benefits in the knowledge base and blog. operationId: post_ai_tasks_create parameters: [] requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/ai_task' responses: '201': description: >- Response returns ID of the created AI task. Using this AI task ID, you can check the status and get the video processing result. Look at GET /ai/results method. content: application/json: schema: $ref: '#/components/schemas/ai_post_response' '400': description: |- Bad request: - "url" is not specified - Queue limit reached (100), try later - etc content: application/json: schema: $ref: '#/components/schemas/streaming_error' '422': description: >- This is advanced functionality; to enable it, contact your manager or the Support Team. content: application/json: schema: $ref: '#/components/schemas/upgraderequired' '429': description: Too many requests, try later content: application/json: schema: $ref: '#/components/schemas/streaming_error' x-codeSamples: - lang: Python source: |- import os from gcore import Gcore client = Gcore( api_key=os.environ.get("GCORE_API_KEY"), # This is the default and can be omitted ) ai_task = client.streaming.ai_tasks.create( task_name="transcription", url="url", ) print(ai_task.task_id) - lang: Go source: "package main\n\nimport (\n\t\"context\"\n\t\"fmt\"\n\n\t\"github.com/G-Core/gcore-go\"\n\t\"github.com/G-Core/gcore-go/option\"\n\t\"github.com/G-Core/gcore-go/streaming\"\n)\n\nfunc main() {\n\tclient := gcore.NewClient(\n\t\toption.WithAPIKey(\"My API Key\"),\n\t)\n\taiTask, err := client.Streaming.AITasks.New(context.TODO(), streaming.AITaskNewParams{\n\t\tTaskName: streaming.AITaskNewParamsTaskNameTranscription,\n\t\tURL: \"url\",\n\t})\n\tif err != nil {\n\t\tpanic(err.Error())\n\t}\n\tfmt.Printf(\"%+v\\n\", aiTask.TaskID)\n}\n" components: schemas: ai_task: type: object properties: task_name: type: string description: Name of the task to be performed enum: - transcription - content-moderation required: - task_name - url example: task_name: transcription|content-moderation url: https://demo-files.gvideo.io/ai_demo_subtitles_nudity_detection.mp4 audio_language: ger category: nsfw allOf: - $ref: '#/components/schemas/ai_transcribe' - $ref: '#/components/schemas/ai_contentmoderation_general' ai_post_response: type: object required: - task_id properties: task_id: type: string format: uuid description: >- ID of the created AI task, from which you can get the execution result example: task_id: aafe70c6-0000-0000-0000-327b65f7670f streaming_error: type: object properties: error: type: string description: Text message with description of error. example: error: Queue limit reached (100), try later. upgraderequired: type: object properties: error: type: string description: >- This is advanced functionality; to enable it, contact your manager or support service. example: error: Feature is disabled. Contact support to enable. ai_transcribe: type: object required: - url - task_name properties: task_name: type: string description: Name of the task to be performed enum: - transcription url: type: string description: >- URL to the MP4 file to analyse. File must be publicly accessible via HTTP/HTTPS. audio_language: type: string description: >- Language in original audio (transcription only). This value is used to determine the language from which to transcribe. If this is not set, the system will run auto language identification and the subtitles will be in the detected language. The method also works based on AI analysis. It's fairly accurate, but if it's wrong, then set the language explicitly. Additionally, when this is not set, we also support recognition of alternate languages in the video (language code-switching). Language is set by 3-letter language code according to ISO-639-2 (bibliographic code). We can process languages: - 'afr': Afrikaans - 'alb': Albanian - 'amh': Amharic - 'ara': Arabic - 'arm': Armenian - 'asm': Assamese - 'aze': Azerbaijani - 'bak': Bashkir - 'baq': Basque - 'bel': Belarusian - 'ben': Bengali - 'bos': Bosnian - 'bre': Breton - 'bul': Bulgarian - 'bur': Myanmar - 'cat': Catalan - 'chi': Chinese - 'cze': Czech - 'dan': Danish - 'dut': Nynorsk - 'eng': English - 'est': Estonian - 'fao': Faroese - 'fin': Finnish - 'fre': French - 'geo': Georgian - 'ger': German - 'glg': Galician - 'gre': Greek - 'guj': Gujarati - 'hat': Haitian creole - 'hau': Hausa - 'haw': Hawaiian - 'heb': Hebrew - 'hin': Hindi - 'hrv': Croatian - 'hun': Hungarian - 'ice': Icelandic - 'ind': Indonesian - 'ita': Italian - 'jav': Javanese - 'jpn': Japanese - 'kan': Kannada - 'kaz': Kazakh - 'khm': Khmer - 'kor': Korean - 'lao': Lao - 'lat': Latin - 'lav': Latvian - 'lin': Lingala - 'lit': Lithuanian - 'ltz': Luxembourgish - 'mac': Macedonian - 'mal': Malayalam - 'mao': Maori - 'mar': Marathi - 'may': Malay - 'mlg': Malagasy - 'mlt': Maltese - 'mon': Mongolian - 'nep': Nepali - 'dut': Dutch - 'nor': Norwegian - 'oci': Occitan - 'pan': Punjabi - 'per': Persian - 'pol': Polish - 'por': Portuguese - 'pus': Pashto - 'rum': Romanian - 'rus': Russian - 'san': Sanskrit - 'sin': Sinhala - 'slo': Slovak - 'slv': Slovenian - 'sna': Shona - 'snd': Sindhi - 'som': Somali - 'spa': Spanish - 'srp': Serbian - 'sun': Sundanese - 'swa': Swahili - 'swe': Swedish - 'tam': Tamil - 'tat': Tatar - 'tel': Telugu - 'tgk': Tajik - 'tgl': Tagalog - 'tha': Thai - 'tib': Tibetan - 'tuk': Turkmen - 'tur': Turkish - 'ukr': Ukrainian - 'urd': Urdu - 'uzb': Uzbek - 'vie': Vietnamese - 'wel': Welsh - 'yid': Yiddish - 'yor': Yoruba default: null subtitles_language: type: string description: >- Indicates which language it is clearly necessary to translate into. If this is not set, the original language will be used from attribute "audio_language". Please note that: - transcription into the original language is a free procedure, - and translation from the original language into any other languages is a "translation" procedure and is paid. More details in [POST /streaming/ai/tasks#transcribe](/docs/api-reference/streaming/ai/create-ai-asr-task). Language is set by 3-letter language code according to ISO-639-2 (bibliographic code). default: null client_user_id: type: string maxLength: 256 default: null description: >- Meta parameter, designed to store your own identifier. Can be used by you to tag requests from different end-users. It is not used in any way in video processing. client_entity_data: type: string maxLength: 4096 default: null description: >- Meta parameter, designed to store your own extra information about a video entity: video source, video id, etc. It is not used in any way in video processing. For example, if an AI-task was created automatically when you uploaded a video with the AI auto-processing option (transcribing, translationing), then the ID of the associated video for which the task was performed will be explicitly indicated here. example: url: https://demo-files.gvideo.io/apidocs/spritefright-blender-cut30sec.mp4 task_name: transcription audio_language: ger ai_contentmoderation_general: type: object required: - url - task_name properties: task_name: type: string description: Name of the task to be performed enum: - content-moderation url: type: string description: >- URL to the MP4 file to analyse. File must be publicly accessible via HTTP/HTTPS. category: type: string description: >- Model for analysis (content-moderation only). Determines what exactly needs to be found in the video. enum: - sport - nsfw - hard_nudity - soft_nudity client_user_id: type: string maxLength: 256 default: null description: >- Meta parameter, designed to store your own identifier. Can be used by you to tag requests from different end-users. It is not used in any way in video processing. client_entity_data: type: string maxLength: 4096 default: null description: >- Meta parameter, designed to store your own extra information about a video entity: video source, video id, etc. It is not used in any way in video processing. For example, if an AI-task was created automatically when you uploaded a video with the AI auto-processing option (nudity detection, etc), then the ID of the associated video for which the task was performed will be explicitly indicated here. example: url: https://demo-files.gvideo.io/ai_demo_subtitles_nudity_detection.mp4 task_name: content-moderation category: nsfw securitySchemes: APIKey: description: >- API key for authentication. Make sure to include the word `apikey`, followed by a single space and then your token. Example: `apikey 1234$abcdef` type: apiKey in: header name: Authorization ````