VOA科学技术2023--谷歌人工智能工具根据文字描述创作音乐

Google AI Tool Creates Music from Written Descriptions

This week, Google researchers published a paper describing results from an artificial intelligence (AI) tool built to create music.

The tool, called MusicLM, is not the first AI music tool to launch. But the examples Google provides demonstrate musical creative ability based on a limited set of descriptive words.

AI shows how complex computer systems have been trained to behave in human-like ways.

Tools like ChatGPT can quickly produce, or generate, written documents that compare well with the work by humans. ChatGPT and similar systems require powerful computers to operate complex machine-learning models. The San Francisco-based company OpenAI launched ChatGPT late last year.

Developers train such systems on huge amounts of data to learn methods for recreating different forms of content. For example, computer-generated content could include written material, design elements, art or music.

ChatGPT has recently received a lot of attention for its ability to generate complex writings and other content from just a simple description in natural language.

Google's MusicLM

Google engineers explain the MusicLM system this way:

First, a user comes up with a word or words that describe the kind of music they want the tool to create.

For example, a user could enter this short phrase into the system: "a continuous calming violin backed by a soft guitar sound." The descriptions entered can include different music styles, instruments or other existing sounds.

Several different music examples produced by MusicLM were published online. Some of the generated music came from just one- or two-word descriptions, such as "jazz," "rock" or "techno." The system created other examples from more detailed¹ descriptions containing whole sentences.

In one example, Google researchers include these instructions to MusicLM: "The main soundtrack of an arcade² game. It is fast-paced and upbeat, with a catchy³ electric guitar riff. The music is repetitive and easy to remember, but with unexpected sounds..."

In the resulting recording⁴, the music seems to keep very close to the description. The team said that the more detailed the description is, the better the system can attempt to produce it.

The MusicLM model operates similarly to the machine-learning systems used by ChatGPT. Such tools can produce human-like results because they are trained on huge amounts of data. Many different materials are fed into the systems to permit them to learn complex skills to create realistic works.

In addition to generating new music from written descriptions, the team said the system can also create examples based on a person's own singing, humming, whistling or playing an instrument.

The researchers said the tool "produces high-quality music...over several minutes, while being faithful to the text conditioning signal."

At this time, the Google team has not released the MusicLM models for public use. This differs from ChatGPT, which was made available online for users to experiment with in November.

However, Google announced it was releasing a "high-quality dataset" of more than 5,500 music-writing pairs prepared by professional musicians called MusicCaps. The researchers took that step to assist in the development of other AI music generators⁵.

The MusicLM researchers said they believe they have designed a new tool to help anyone quickly and easily create high-quality music selections. However, the team said it also recognizes some risks linked to the machine learning process.

One of the biggest issues the researchers identified was "biases⁷ present in the training data." A bias⁶ might be including too much of one side and not enough of the other. The researchers said this raises a question "about appropriateness for music generation for cultures underrepresented in the training data."

The team said it plans to continue to study any system results that could be considered cultural appropriation⁸. The goal would be to limit biases through more development and testing.

In addition, the researchers said they plan to keep improving the system to include lyrics⁹ generation, text conditioning and better voice and music quality.

Words in This Story

artificial intelligence – n. the development of computer systems that have the ability to perform work that normally requires human intelligence

style –n. a particular form or design, usually used in comparing forms of art or handiwork

instruction –n. a description of how to do something

arcade – n. an area containing many electronic and other coin-operated games

upbeat – adj. full of hope and happiness

repetitive – adj. saying or doing something over and over again

hum – v. to make a musical sound without opening your mouth

whistle – v. to make a high sound by forcing air through a small hole in the mouth

faithful – adj. staying firm about an idea or belief

appropriate – adj. the level to which something is right for a situation

cultural appropriation – n. when members of a culture in a society, often the main culture, use a practice of another, often minority, culture, without fully¹⁰ understanding the meaning or importance of the practice.

1 detailed   

adj.详细的，详尽的，极注意细节的，完全的

参考例句：

• He had made a detailed study of the terrain.他对地形作了缜密的研究。
• A detailed list of our publications is available on request.我们的出版物有一份详细的目录备索。

2 arcade   

n.拱廊；(一侧或两侧有商店的)通道

参考例句：

• At this time of the morning,the arcade was almost empty.在早晨的这个时候，拱廊街上几乎空无一人。
• In our shopping arcade,you can find different kinds of souvenir.在我们的拱廊市场，你可以发现许多的纪念品。

3 catchy   

adj.易记住的，诡诈的，易使人上当的

参考例句：

• We need a new slogan.The old one's not catchy enough.我们需要新的口号，旧的不够吸引人。
• The chorus is very catchy to say the least.副歌部分很容易上口。

4 recording   

n.录音，记录

参考例句：

• How long will the recording of the song take?录下这首歌得花多少时间？
• I want to play you a recording of the rehearsal.我想给你放一下彩排的录像。

5 generators   

n.发电机，发生器( generator的名词复数 )；电力公司

参考例句：

• The factory's emergency generators were used during the power cut. 工厂应急发电机在停电期间用上了。
• Power can be fed from wind generators into the electricity grid system. 电力可以从风力发电机流入输电网。 来自《简明英汉词典》

6 bias   

n.偏见，偏心，偏袒；vt.使有偏见

参考例句：

• They are accusing the teacher of political bias in his marking.他们在指控那名教师打分数有政治偏见。
• He had a bias toward the plan.他对这项计划有偏见。

7 biases   

偏见( bias的名词复数 )； 偏爱； 特殊能力； 斜纹

参考例句：

• Stereotypes represent designer or researcher biases and assumptions, rather than factual data. 它代表设计师或者研究者的偏见和假设，而不是实际的数据。 来自About Face 3交互设计精髓
• The net effect of biases on international comparisons is easily summarized. 偏差对国际比较的基本影响容易概括。

8 appropriation   

n.拨款，批准支出

参考例句：

• Our government made an appropriation for the project.我们的政府为那个工程拨出一笔款项。
• The council could note an annual appropriation for this service.议会可以为这项服务表决给他一笔常年经费。

9 lyrics   

n.歌词

参考例句：

• music and lyrics by Rodgers and Hart 由罗杰斯和哈特作词作曲
• The book contains lyrics and guitar tablatures for over 100 songs. 这本书有100多首歌的歌词和吉他奏法谱。

10 fully   

adv.完全地，全部地，彻底地；充分地

参考例句：

• The doctor asked me to breathe in,then to breathe out fully.医生让我先吸气，然后全部呼出。
• They soon became fully integrated into the local community.他们很快就完全融入了当地人的圈子。