Smart Music

Media Lab's Smart Music combines music theory knowledge with deep learning technology to deliver intelligent music analysis, music creation, and a fusion of musical operations. The capabilities include music labeling, climax detection, music structure analysis, music transcription, singing scoring, intelligent composition, video soundtrack, and other functions.

Music Scoring

Music Scoring is a rating solution to judge singing performances based on pitch, rhythm, and other dimensions.

Scenarios

  • Karaoke
    Automatic scoring of karaoke scenes
  • Music Teaching
    Multi-dimensional analysis and scoring of music

Music Labeling

Music Labeling analyzes music to understand its various elements such as intensity, mood, style, emotion, and other dimensions of music understanding. After analysis, the solution automatically generates labels for the music such as style, mood, intensity, presence of human vocals, instrumental music only, and more. There are currently more than 20 categories that the solution can label.

Music
StyleEmotionStrengthVocal
Instrumental MusicPeacefulLeisurelyNon-Exist

Scenarios

  • Music/Dance Teaching
    Perform multi-dimensional analysis of music, provide assistance for music and dance teaching, and automatically score music according to tags
  • Music/Ads Recommendation
    Music media asset management, adding tags to music, and combining with other portrait expressions can be used as recommendation basis
  • Scenic/park interactive experience area
    Customized Pan-entertainment Interactive Mini Games

Prelude Skipping

Prelude Skipping scans music and identifies the timestamp when the intro (prelude music) ends and the singing begins.

Scenarios

  • Karaoke
    Skip the song intros to avoid waiting and start singing the lyrics.
  • Cinema/Museum Interactive Experience Area
    Dubbing movies or videos for interactive activities or games.

Drumbeat Recognition

Drumbeat Recognition automates the process of scanning music to locate the time positions of the drum beats to determine the rhythm. Use this capability to quickly sync video tracks to the beat positions or use the recognized rhythm to generate drum beats to the music.

Scenarios

  • Audio/Video Clips
    Music clipping by drumbeats
  • Music automatic drummer

Pitch Recognition

Pitch Recognition scans audio, whether vocal singing or musical instrument performance, to calculate in real-time the audio pitch. This solution also supports midi output for the audio to be used on mobile phones.

Original
Pitch Result

Scenarios

  • Music Teaching and Karaoke Entertainment
    Music scoring, visualization of singing, singing coaching
  • Cinema/Museum Interactive Experience Area
    Dubbing movies or videos for interactive activities or games.

Highlights Recognition

Highlights Recognition scans music to detect the climatic, emotional, or chorus highlights of a song. Users can skip directly to song highlights in karaoke apps. Editors can quickly create music highlights clips. Short video content creators can access a catalog of music highlights to complement their videos.

Scenarios

  • Karaoke
    Singing song highlights
  • Audio Clip
    Music highlights clip, music highlights library

Source Separation

Source Separation takes complete audio tracks and separates the audio into distinct sources such as vocals, accompaniment, guitar, piano, drums, bass, and more. Scenarios to use audio separation include advertisements, games, marketing, and more.

Scenarios

  • Video Soundtrack
    The original audio needs to be removed or only the vocal dialogue is retained during the video soundtrack
  • TV Show or Film Dubbing
    For secondary dubbing of film and television works, the original sound needs to be separated and removed

Audio Denoising

Audio Denoising analyzes input audio to remove noisy background sounds to return the audio to a clear and polished state. Applications for this capability are to remove noise captured unintentionally during the filming process. Another application is to clean up the audio for further processing downstream.

Scenarios

  • Audio Editing
    Audio background noise removal
  • Audio Creation
    Secondary music creation tool

Sheet Music Conversion

Sheet Music Conversion analyzes piano performance music and identifies the instrument pitch and notes. The solution can then transcribe the notes to midi or sheet music files for sheet music libraries or provides sheet music for personal analysis and study.

Canon

Scenarios

  • Automatic sheet music generation
  • Music teaching aid

Instrument Conversion

Instrument Conversion uses deep learning for music understanding and identifies the instruments and notes used in the music composition. The music can then be converted to playback using one of 31 possible instruments from our API.

Scenarios

  • Music generation
  • Background music production

Style Conversion

Style Conversion analyzes the input music to determine the composition and instruments used. The solution can then perform conversion with instrument substitution to fit a style selected by the user. Styles of music supported include opera, Chinese style, pop, ballad, sentimental, rock, electronic, R&B, and more.

Scenarios

  • Music generation
  • Background music production

Vocal Transcription

Input the original music, strip the vocal part, calculate the pitch and interval of each note, and output a standard MIDI file.

Original
MIDI

Scenarios

  • Automatic MIDI generation
  • Singing score

Smart Composer

Just enter a few notes or cues to automatically compose single-track music.

Scenarios

  • Automatic generation of music material
  • Solve BGM copyright issues

Smart Soundtrack

Input the video, automatically match the background music in the music library that matches the video content and atmosphere, and generate a video with BGM without manual editing. Support custom music library. It includes audio and video capabilities such as automatic score, music MV, etc., and supports access to the audio and video library of the business team, which can speed up content production and accelerate content distribution and dissemination.

Smart Soundtrack

Scenarios

  • Short Video
    Provide creators with recommended music, or automatically generate soundtracks to accelerate content production
  • Long Video
    Automatically match bgm for video clips such as film and television dramas, Vlogs, etc., to assist video editing, the original audio needs to be removed or only the vocal dialogue should be retained during soundtracking
  • Exhibition/Scenic Spot
    Low-cost automatic generation of promotional videos

Music Movie

Import the original movie, and automatically edit the best movie clips from the spatial and temporal dimensions, and match the climax of the music.

Scenarios

  • The player loops
  • Video generation

Music Video

Enter the original music, recommend the most matching material in the video library for synthesis, find the video that best expresses its mood or rhythm for the music, and automatically make a wonderful music MV. Support custom video library to create exclusive MV. It includes audio and video capabilities such as automatic score, music MV, etc., and supports access to the audio and video library of the business team, which can speed up content production and accelerate content distribution and dissemination.

Output Video

Scenarios

  • Music products
    Make music into MV to increase the content dimension of music products
  • Smart Scenic Area/Smart Park/Amusement Park
    A collection of wonderful videos of the park/recreation project for playback on the big screen/official website

Music Vlog

Identify the clips with richer content and more aesthetic features in the video, and automatically complete the editing and soundtrack work, and quickly generate Vlogs. It includes audio and video capabilities such as automatic score, music MV, etc., and supports access to the audio and video library of the business team, which can speed up content production and accelerate content distribution and dissemination.

Scenarios

  • Supports general scenarios, and has been optimized for specific scenarios such as sports, animation, pan-entertainment, etc.
In need of demos or documentations?
Tencent Media Lab
/
We would like to use performance and analytics cookies (“Cookies”) to help us recognize whether you are a returning visitor and to track the number of website views and visits. For more information about the Cookies we use and your options (including how to change your preferences) see our Cookies Policy.