All Products
Search

This Product

    ApsaraVideo Media Processing:Features

    Document Center

    ApsaraVideo Media Processing:Features

    Last Updated:Jun 02, 2026

    ApsaraVideo Media Processing (MPS) converts media files to suit different network bandwidths, devices, and user needs. It analyzes media content, text, speech, and scenes to provide automated review, content recognition, and smart editing.

    Audio and video transcoding

    Transcoding converts media streams into different resolutions, encoding formats, or container formats for various network bandwidths and playback devices. MPS supports mainstream encoding and container formats, and allows adding watermarks, captions, or simple edits during transcoding. For features unavailable in the console or API, contact your account manager.

    Important

    To use the following features, you must submit a transcoding job. Standard transcoding fees apply, based on the specifications and duration of the output video. For billing details, see Transcoding Pricing.

    Parameter

    Type

    Description

    Input file

    Container format

    • Video: 3GP, AVI, FLV, MP4, M3U8, MPG, ASF, WMV, MKV, MOV, TS, WebM, MXF, and VOB.

    • Audio: AAC, FLAC, M4A, MP3, MP4, and OGG.

    • Caption: ASS, SSA, SRT, and VTT.

    Video encoding format

    Apple ProRes, AVS+, AVS, AVS2, H.263, H.263+, H.264/AVC, H.265/HEVC, H.266/VVC, MJPEG, MPEG-1, MPEG-2, MPEG-4, QuickTime, RealVideo, VP8, VP9, and Windows Media Video.

    Audio encoding format

    AAC, AC3, ADPCM, AMR, DSD, EAC3, MP1, MP2, MP3, PCM, RealAudio, Vorbis, and Windows Media Audio.

    File size

    Up to 100 GB.

    Chroma subsampling

    Examples: 4:2:2, 4:2:0.

    Output file

    Container format
    Note

    • Container formats must be paired with compatible encoding formats, as listed in Supported formats. For features unavailable in the console or API, contact your account manager.

    • Remuxing changes only the container format; it does not re-encode the audio or video streams. Supported output formats for remuxing are MP4, HLS, and FLV.

    • Video: HLS, DASH, CMAF, 3GP, AVI, FLV, F4V, fMP4, MKV, MOV, MP4, MPEG, TS, MXF, and WebM.

    • Animated image: GIF and WEBP.

    • Audio: AAC, M4A, MP2, MP3, MP4, OGG, FLAC, and WAV.

    Encoding format

    • Video: H.263, H.264/AVC, H.265/HEVC, H.266/VVC, VP8, VP9, AV1, AVC-Intra, AVS2, MPEG-1, MPEG-2, MPEG-2 422, MPEG-4, and Apple ProRes.

    • Animated image: GIF and WEBP.

    • Audio: AAC, AC3, EAC3, MP2, MP3, FLAC, OPUS, VORBIS, Windows Media Audio, and pcm_s16le.

    Encoding profile

    • H.264 supports Baseline, Main, and High encoding profiles.

    • AAC supports aac_low, aac_he, aac_he_v2, aac_ld, and aac_eld encoding profiles.

    Resolution

    • H.264 output resolution: [128, 4096] pixels.

    • H.265, H.266, and AV1 output resolution: [128, 8192] pixels.

    Bitrate

    • H.264 output bitrate: [10, 50000] Kbps.

    • H.265, H.266, and AV1 output bitrate: [10, 200000] Kbps.

    Frame rate

    The output frame rate range is (0, 60] fps.

    Bit depth

    • H.264 supports 8-bit.

    • H.265 supports up to 12-bit.

    Pixel format

    yuv420p, yuvj420p, yuv422p, yuvj422p, yuv444p, yuvj444p, etc.

    Bitrate control

    VBR (Variable Bitrate), CBR (Constant Bitrate), ABR (Average Bitrate), and CRF (Constant Rate Factor).

    Scan mode

    Supported modes include the original video mode, automatic deinterlacing, interlaced scan, and progressive scan.

    Narrowband HDTM

    Narrowband HDTM uses Alibaba Cloud's transcoding technology to improve video compression, reducing file sizes, minimizing playback buffering, and lowering storage and bandwidth costs.

    Important

    To use these features, you must select the corresponding Narrowband HD transcoding template when you submit a transcoding job. You are charged for Narrowband HD transcoding based on the specifications and duration of the output video.

    Type

    Description

    Narrowband HDTM 1.0

    Analyzes scenes, motion, content, and textures to reduce bitrate by 20%–40% while maintaining quality, or deliver higher definition at the same bandwidth. Supports H.264 and H.265. Other parameters match standard transcoding. Start a free trial.

    Narrowband HDTM 2.0

    Surpasses standard encoders by integrating lightweight definition restoration and quality enhancement, reducing bitrate by 40%–60% while maintaining quality or delivering premium HD on low-bandwidth networks. Supports H.264 and H.265. Other parameters match standard transcoding. Start a free trial.

    Audio and enhancement

    Audio enhancement

    ApsaraVideo Audio Lab combines signal processing with deep learning to enhance and restore audio across various scenarios.

    Important

    To use the following features, you must configure the corresponding parameters when you submit a transcoding job. The audio portion incurs audio and video enhancement fees based on the output audio's specifications and duration. The video portion incurs transcoding fees based on the services used. To enable an audio enhancement transcoding template, contact Alibaba Cloud technical support by joining DingTalk group 32171220.

    Feature

    Description

    Sound enhancement

    Enhances mono, stereo, 5.1, and 7.1 audio to deliver cinematic-quality sound across headphones and speakers. Ideal for music, speeches, and film.

    Volume normalization

    Adjusts and normalizes loudness to resolve volume fluctuations from inconsistent sources in short video feeds and continuous music playback.

    High-speed transcoding

    High-speed transcoding splits a video into segments and transcodes them in parallel, boosting speed 5–30x. Ideal for time-sensitive content such as news, sports events, and premieres.

    Important

    To accelerate a transcoding job, submit it to a high-speed transcoding queue. In addition to the standard fees for audio and video transcoding or audio-visual enhancement, an extra fee applies for high-speed transcoding. This fee is based on the output video's specifications, duration, and the actual speed boost.

    Type

    Description

    Speed boost

    The speed boost is typically 5 to 30 times, depending on the input video. You can set an expected speed boost on the high-speed transcoding queue, such as 5x, 10x, 20x, or 30x.

    Applicable videos

    Recommended for videos longer than 30 minutes or those requiring complex processing such as high frame rates, UHD resolutions, or audio-visual enhancement. Limits on high-speed transcoding.

    Fallback policy

    Segmentation is not supported for all videos. If a video that is incompatible with high-speed transcoding is submitted to a high-speed transcoding queue, the job automatically falls back to the regular transcoding mode.

    Additional processing

    Media information

    Retrieve metadata from audio and video files in OSS, such as resolution, bitrate, frame rate, codec, and file format.

    Important

    To use this feature, you must submit a media information job. This feature is billed for API calls based on the number of requests.

    Simple editing

    Perform simple video editing, such as audio extraction, video extraction, merging, clipping, and audio mixing.

    Important

    To use the following features, you must set the corresponding parameters when you submit a transcoding job. This feature incurs transcoding fees based on the specifications and duration of the output video.

    Feature

    Description

    Related API parameters

    Console operation

    Audio extraction

    Extracts the audio track from a video file by disabling the video track.

    Remove

    Supported

    Video extraction

    Extracts the video track from a video file by disabling the audio track.

    Remove

    Supported

    Black bar removal

    Detects and automatically crops black bars from the video frame.

    Crop

    Not supported

    Video cropping

    Specifies the width and height of the area to retain after cropping, and the position of this area relative to the top-left corner of the input video frame. The surrounding areas are automatically cropped.

    Crop

    Not supported

    Black bar addition

    Specifies the width and height of the output video after black bars are added, and the position of the input video frame relative to the top-left corner of the output frame. The surrounding gaps are filled with black bars.

    Pad

    Not supported

    Orientation adaptation

    Defines output resolution by long and short sides instead of width and height. Recommended when input videos include both landscape and portrait orientations.

    LongShortMode

    Supported

    Video rotation

    Sets the rotation angle of the video.

    Rotate

    Supported

    Video merging

    Merge up to 100 videos. You can set the start time and duration for each video to be merged.

    MergeList or MergeConfigUrl

    Not supported

    Video clipping

    • Video segment clipping: Extracts a video segment of a specified duration starting from a specified time point.

    • End trimming: Trims a segment of a specified duration from the end of the video.

    Clip

    Supported

    Opening and tail slates

    Overlays a dynamic logo at the beginning and appends tail slate content for brand recognition and copyright display.

    OpeningList and TailSlateList

    Supports tail slate addition

    Blurring

    Blurs a specified area within the video.

    DeWatermark

    Not supported

    Audio mixing

    Merges two audio tracks, which is ideal for adding background music.

    Amix

    Not supported

    Video snapshots

    Take snapshots of a specified size at specified time points in a video. You can use snapshots for video covers, sprites, and thumbnails on player progress bars.

    Important

    To use the following features, you must submit a snapshot job. This feature is billed for API calls based on the number of snapshots taken.

    Feature

    Description

    Related API parameters

    Console operation

    Static snapshot

    Takes JPG snapshots of a specified size at specified time points in a video. The following sampling methods are available:

    • Single snapshot: Takes one snapshot at a specified time point. This method supports synchronous and asynchronous calls.

    • Interval snapshot: Takes snapshots at a specified interval from a start time. Stops when the count is reached or the video ends. Interval is in seconds. Supports only asynchronous calls.

    • Average snapshot: Takes a specified number of snapshots at regular intervals from a specified time point to the end of the video. This method supports only asynchronous calls.

    • Time-point snapshot: Takes snapshots at a specified set of time points. This method supports only asynchronous calls.

    SnapshotConfig

    Supported

    Sprite snapshot

    Stitches static snapshots into a single large image (sprite) based on layout rules. Output is in JPG format. Supports only asynchronous calls. A single sprite request retrieves multiple images, reducing request count and improving client performance.

    TileOut, TileOutputFile

    Not supported

    WebVTT snapshot

    Generates a VTT file for static snapshots or a sprite, containing timestamps, file URLs, and coordinate information. To display an image, retrieve and parse the VTT file first. Useful for player progress bar thumbnails.

    SubOut

    Supported

    Keyframe snapshot

    This feature takes snapshots only at keyframes. If a specified time point is not a keyframe, the service uses the nearest keyframe instead.

    FrameType

    Supported

    First-frame black screen detection

    You can enable black screen detection for the first frame (time=0). A black screen is defined by the percentage of black pixels and a color value threshold. The service scans the first 5 seconds: if a non-black frame is found, it is captured. Otherwise, a single-snapshot job fails; a multi-snapshot job captures the first black frame.

    BlackLevel, PixelBlackThreshold

    Supported

    Video watermarks (visible watermarks)

    Adds visible watermarks, such as a company logo or TV station icon, to a video to enhance brand identity, highlight copyright, and increase product recognition. To add invisible blind watermarks for copyright tracing, see the Digital watermarking (blind watermarks) section below.

    Important

    To use the following features, you must submit a transcoding job and specify the watermark assets and an optional watermark template. This feature incurs transcoding fees based on the specifications and duration of the output video, and is billed for API calls based on the number of watermarks.

    Feature

    Description

    Related API parameters

    Console operation

    Image watermark

    • You can add up to four watermarks to a single video. You can set the start time, duration, position, and size for each watermark.

    • Supported image watermark formats include static PNG (.png), animated PNG (.apng), MOV (.mov), and GIF (.gif).

    • You can use a watermark template to simplify development.

    WaterMarks

    Supported

    Text watermark

    • You can add up to four watermarks to a single video. You can set the start time, duration, position, and size for each watermark.

    • You can configure the text content, font size, font, color, opacity, and outline effects. For supported fonts and colors, see Fonts and Color.

    • Text watermarks do not support watermark templates.

    WaterMarks

    Not supported

    Subtitles

    Add subtitles to videos for multi-language support or for viewing in silent or noisy environments. This feature also makes video content accessible to users with hearing impairments.

    Important

    To use the following features, you must submit a transcoding job or initiate transcoding through a workflow. This feature incurs transcoding fees based on the specifications and duration of the output video.

    Feature

    Description

    Related API parameters

    Console operation

    Caption packaging

    Uses a workflow to package subtitle files with audio and video streams into a master playlist (.m3u8 or .mpd). Supports up to four subtitle tracks for in-player switching. Formats: VTT for HLS; VTT, STL, and TTML for DASH.

    • HLS packaging: ExtXMedia

    • DASH packaging: inputconfig

    Supported

    Video packaging

    Packaging is the process of generating a master playlist file from multiple bitrate renditions, audio tracks, and subtitle tracks. In streaming media scenarios, this enables:

    • Adaptive bitrate streaming: Adapts the stream quality to the user's network conditions to reduce startup latency and playback buffering.

    • Ad insertion: Flexibly inserts video ads between segments.

    Important

    To use the following features, you must initiate transcoding through a workflow. This feature incurs transcoding fees based on the specifications and duration of the output video.

    Feature

    Description

    Related API parameters

    Console operation

    HLS protocol

    Packages content using the Apple HLS protocol. It supports a two-level index structure. The index file is in M3U8 format and the video file is in TS format.

    See How to package content in HLS.

    Supported

    CMAF protocol

    Packages content using the CMAF protocol. It supports output in HLS or DASH format.

    None

    Not supported

    Custom segment duration

    Specifies up to 10 segment time points and a separate interval for the remaining duration. The interval can range from 1 to 60 seconds. This feature helps you set the segment duration based on the client's bandwidth conditions to reduce the time to first frame.

    Segment

    Not supported

    Video encryption

    Important

    To use the following features, you must initiate transcoding through a workflow. This feature incurs transcoding fees based on the specifications and duration of the output video.

    Feature

    Description

    Related API parameters

    Console operation

    HLS encryption

    Encrypts video content using HLS AES-128 with self-managed keys or KMS. Decrypts and plays on any standard HLS-compliant player. Provides strong security with broad device compatibility.

    Not applicable

    Supported

    Alibaba Cloud proprietary cryptography

    Encrypts video files using Alibaba Cloud's proprietary protocol into encrypted HLS format with KMS key management. Playback requires ApsaraVideo Player, preventing playback and redistribution even if files are downloaded. Ideal for online education and premium content.

    Not applicable

    Supported

    Video AI

    Automated review

    This feature reviews media content—including titles, descriptions, covers, videos, and audio—to detect policy violations. Common use cases include short video platforms, live streaming, and media compliance.

    Important

    To use the following features, you must submit a media review job. You are charged for automated review based on the duration of the successfully processed video.

    Feature

    Content

    Description

    Content moderation

    Pornography detection

    Detects pornographic and sexually suggestive content in audio, text, and visuals.

    Terroristic content detection

    Detects terroristic and politically sensitive content across more than 10 categories, including weapons, gore, specific attire, explosions, special symbols, crowds, and parades.

    Ad and QR code detection

    Detects various forms of ads in media files, including text, watermarks, QR codes, spammy ads, and mini program codes.

    Logo detection

    Detects various logos in videos and images, such as TV station logos, trademarks, and watermarks. This is often used for copyright protection.

    Undesirable scene detection

    Detects undesirable scenes that require moderation, such as picture-in-picture (PiP), smoking, in-car live streaming, and meaningless footage.

    Audio anti-spam

    Detects policy violations in audio, such as pornographic, terroristic, and abusive content. Supports both Chinese and English.

    Media fingerprinting

    This feature uses Alibaba Cloud's technology to create a unique fingerprint for a media file. By extracting and comparing these fingerprints for videos, images, audio, and text, you can find duplicate content and trace the source of media clips..

    Important

    To use the following features, you must submit a media fingerprinting job. You are charged for media fingerprinting based on the duration of the successfully processed video or audio.

    Feature

    Description

    Video fingerprinting

    Extracts a visual fingerprint from a video. You can add the fingerprint to a library for analysis and comparison to find similar videos.

    Audio fingerprinting

    Extracts an audio fingerprint. You can add the fingerprint to a library for analysis and comparison to find similar audio tracks.

    Image fingerprinting

    Extracts a visual fingerprint from an image. You can add the fingerprint to a library for analysis and comparison to find similar images.

    Text fingerprinting

    Extracts a fingerprint from text content. You can add the fingerprint to a library for analysis and comparison to find similar text.

    Service management

    Category

    Description

    API

    Console

    Media management

    Upload, manage, and publish media files.

    N/A

    N/A

    Workflow orchestration

    Automates cloud-based media processing by automatically starting jobs when a media file is uploaded.

    N/A

    Supported

    Transcoding template

    A transcoding template defines a reusable set of transcoding parameters, simplifying job creation and workflow configuration. Templates are custom, customized, or system-preset.

    TemplateId

    Supported

    Watermark template

    A watermark template contains parameters that define a watermark's properties, such as its position and size. Using a template simplifies the process of adding watermarks.

    WaterMarkTemplateId

    Supported

    Transcoding priority

    Sets the priority of a transcoding job within an MPS queue. You can define up to 10 priority levels.

    Priority

    Not supported

    Conditional transcoding

    Compares source video properties (bitrate, resolution, audio bitrate) against output settings. If source values are lower, the job is skipped or the source is passed through unchanged.

    Examples: IsCheckReso and IsCheckResoFail

    Supported

    MPS queue

    An MPS queue holds asynchronous jobs (transcoding, snapshotting, etc.) for scheduling and execution. You can create multiple queues with up to 10 priority levels per queue.

    Priority

    Not supported

    Message notification

    Asynchronous jobs run in the background. Bind an MNS topic or queue to an MPS queue or workflow to receive automatic notifications when jobs complete or workflows start or finish.

    NotifyConfig

    Supported