Tencent Unveils Qinle AI Music Composition Model, Democratizing MusicCreation
Shenzhen, China – Tencent, the Chinese tech giant,has announced the launch of Qinle, an AI music composition model designed to empower both amateur and professional musicians. Developed jointly by Tencent AI Lab and Tencent MusicEntertainment (TME) Tianqin Lab, Qinle utilizes advanced deep learning techniques to generate original music based on user input.
The model, currently availableon Tencent Music’s 启明星 (Qimingxing) platform, allows users to create music by providing keywords, descriptive sentences, or even audio samples. Qinle then generates a complete musical composition, including multi-track scores,complete with melody, harmony, accompaniment, and percussion.
Beyond Simple Generation: Qinle Offers Advanced Editing Capabilities
Qinle goes beyond simple music generation, offering a suite of editing tools for fine-tuning the created music.Users can seamlessly extend existing compositions, re-generate specific sections, adjust instrumentation, and modify tempo and rhythm. This level of control empowers users to refine their musical ideas and achieve their desired sonic landscapes.
Qinle’s Technical Underpinnings
At the heart of Qinle lies a sophisticated architecture built uponseveral key components:
- Audio-Text Alignment Model: This module employs contrastive learning to establish a shared feature space between audio and textual descriptions. This allows Qinle to understand the semantic relationships between text and sound, enabling it to generate music that aligns with user input.
- Score/Audio Representation Extraction: Qinle converts scores or audio into sequences of discrete features (tokens). These tokens represent the musical elements and provide the foundation for the large language model’s predictions.
- Large Language Model: Utilizing a decoder-only structure, the large language model predicts the next token based on the input sequence. This predictivecapability enables the generation of continuous musical elements, effectively translating text into music.
- Flow Matching and Vocoder Technology: To convert the predicted audio feature sequences into audible sound, Qinle employs flow matching and vocoder technology, enhancing the realism and quality of the generated audio.
A Future with Enhanced Musical Expression
The Qinle team is actively working to enhance the model’s capabilities, aiming to incorporate vocal generation and lyric writing in the future. This will further expand the creative possibilities for musicians, allowing them to express themselves through a wider range of musical elements.
Accessibility and Impact
Qinle’s availabilityon the Qimingxing platform makes it accessible to a broad audience, fostering a more inclusive and democratized music creation environment. This empowers individuals with limited musical training to explore their creativity and realize their musical visions.
The introduction of Qinle marks a significant step in the evolution of AI-powered music creation. Itoffers a powerful tool for both seasoned composers and aspiring musicians, pushing the boundaries of musical expression and inspiring a new generation of creators.
【source】https://ai-bot.cn/tencent-ai-music-llm/
Views: 1