Intelligent Media Services:SubmitIProductionJob

Input and output fields

Cover

Input: a video file. Output: multiple images (three by default, which must be distinguished using placeholders). The output format is PNG for static images or GIF for animated images, depending on the settings in JobParams.

VideoDelogo

Input: a video file. Output: a video in MP4 format with the logo removed.

VideoDetext

Input: a video file. Output: a video in MP4 format with captions removed.

CaptionExtraction

Input: a video file. Output: a caption file in SRT format.

HybridCaptionExtraction

Input: a video file. Output: a caption file in SRT format.

VideoGreenScreenMatting

Input: a video file. Output: a video with the green screen background removed. The format is MP4 or WebM, depending on the settings in JobParams.

FaceBeauty

Input: a video file. Output: a beautified video in MP4 format.

VideoH2V

Input: a video file. Output: a video in MP4 format converted from a horizontal to a vertical aspect ratio.

MusicSegmentDetect

Input: an audio file. Output: a JSON file containing the chorus detection results.

AudioBeatDetection

Input: an audio file. Output: a JSON file containing the beat detection results.

AudioQualityAssessment

Input: an audio file. No output file is generated. The audio quality assessment results are returned directly in the response of the QueryIProductionJob operation.

SpeechDenoise

Input: an audio file. Output: a noise-reduced audio file in WAV format.

AudioMixing

Input: an audio file. Output: a mixed audio file in WAV format. For details on how to specify additional audio files for mixing, see the JobParams parameters below.

MusicDemix

Input: an audio file (typically a song). Output: two audio files resulting from source separation. You must include the {resultType} placeholder in the output path to distinguish between the vocals and the accompaniment.

JobParams JSON fields

Cover

• Model: string. The model for the smart cover. If this parameter is left empty, a static image is generated. If set to gif, an animated image is generated.

VideoDelogo

• LogoModel: string. The type of station logo to remove. Valid values are tv (for television station logos) and internet (for online media logos). You can specify multiple values, separated by commas.

• Boxes: string. The bounding boxes for the target logos. The coordinates are normalized values relative to the top-left corner of the video, in the format [xmin, ymin, width, height]. Supports up to two bounding boxes. Example: "[[0, 0, 0.3, 0.3], [0.7, 0, 0.3, 0.3]]".

VideoDetext

• LimitRegion: list. Specifies the region(s) for caption detection. The coordinates are normalized values relative to the top-left corner, specified as [xmin, ymin, width, height]. You can specify multiple detection regions. Example: [[0, 0, 0.3, 0.3], [0.7, 0, 0.3, 0.3]]. Note: If this parameter is not set, the default detection region is the bottom 30% of the video.

• Time: list. The time range for caption removal, specified in seconds as [start_time, end_time]. For example, [5, 20] removes captions between the 5-second and 20-second marks of the video.

  • The Time parameter can be a one-dimensional array, such as [5, 20], to specify a single time range.

  • The Time parameter can also be a two-dimensional array, such as [[5, 20], [25, 43], [51, 80]], to specify multiple time ranges (supported only when modelId is set to algo-video-detext-new).

CaptionExtraction

• fps: integer (Optional). The sampling frame rate. Range: [2, 10]. Default: 5.

• roi: list. The region of interest (ROI) for caption extraction. Only captions within this region are extracted. The format is [[top, bottom], [left, right]], using normalized values. For example, [[0.5, 1], [0, 1]] specifies the bottom half of the video. If this parameter is not provided, the default region is the bottom 1/4 of the video.

• lang: string. The recognition language. Valid values: ch (Chinese), en (English), and ch_ml (Chinese-English mixed). Default: ch.

• track: string. If set to main, only the main caption track is extracted. If this parameter is not set, the system extracts all captions that appear in the specified region by default.

HybridCaptionExtraction

• fps: integer (Optional). The sampling frame rate. Range: [2, 10]. Default: 5.

• roi: list. The bounding box of the target caption, in the format [bx, by, bw, bh]. If this parameter is not provided, the default region is the bottom 1/4 of the video.

  • bx: The normalized x-coordinate of the top-left corner of the bounding box, relative to the video width. Example: 0.1.

  • by: The normalized y-coordinate of the top-left corner of the bounding box, relative to the video height. Example: 0.0.

  • bw: The normalized width of the bounding box, relative to the video width. Example: 0.3.

  • bh: The normalized height of the bounding box, relative to the video height. Example: 0.2.

• lang: string. The recognition language. Valid values: zh (Chinese) and en (English). Default: zh.

• track: string. If set to main, only the main caption track is extracted. If this parameter is not set, the system extracts all captions that appear in the specified region by default.

VideoGreenScreenMatting

• bgimage: string. The background image to replace the green screen. Example: http://example-image-****.example-location.aliyuncs.com/example/example.jpg. If you do not set this parameter, the service outputs a WebM video with an alpha channel.

FaceBeauty

• beauty_params: string. The beautification parameters. Example: "whiten=20,smooth=50,face_thin=50". For more information, see Parameter field descriptions.

VideoH2V

None

MusicSegmentDetect

None

AudioBeatDetection

None

AudioQualityAssessment

None

SpeechDenoise

Input audio requirements: The audio file must be in WAV format with a sample rate of 16 kHz or 48 kHz.

AudioMixing

• inputs: list. A list of URLs for the audio tracks to mix. Currently, only one audio track is supported. Example: {"file":"http://example-bucket-****.oss-cn-shanghai.aliyuncs.com/2.mp4"}

MusicDemix

None