📚 Year 12 Edexcel Computer Science: Speech & Audio Revision Guide | 口语/听力备考专项
Computers are not just about silent code – modern systems regularly process, store, and produce sound, from simple beeps to complex voice commands. For Year 12 Edexcel Computer Science students, mastering the principles of audio representation and speech technology is a subtle but crucial part of the syllabus. This guide turns the abstract notion of “speaking and listening” in computing into clear, exam-ready concepts, covering everything from sampling to compression, and from MIDI to voice recognition. Each section links directly to Edexcel AS-level expectations, giving you the tools to tackle data representation, storage calculations, and application scenarios with confidence.
计算机并不只是沉默的代码——现代系统经常处理、存储并产生声音，从简单的提示音到复杂的语音指令。对于 Year 12 Edexcel 计算机科学的学生来说，掌握音频表示和语音技术的原理是课程中微妙但关键的一部分。这份指南将计算机领域里“说与听”的抽象概念转化为清晰、适用于考试的知识点，涵盖从采样到压缩、从 MIDI 到语音识别的所有内容。每一部分都直接对应 Edexcel AS 级别的要求，让你能够从容应对数据表示、存储容量计算和应用情境题。
1. Sound as Data | 把声音看作数据
Sound is an analogue phenomenon – a continuous wave of pressure variations travelling through a medium. Computers, however, work with discrete binary values. To make sound usable in a computer system, we must convert it from a smooth, continuous signal into a stream of numbers. This process sits at the heart of the data representation topic in Edexcel’s specification, and it explains why we need sampling, quantisation, and encoding. Understanding that sound is captured and stored as numerical data helps you see the direct link between perceived audio quality and the binary file stored on disk.
声音是一种模拟现象——连续的压力变化波在介质中传播。但计算机只能处理离散的二进制值。要让声音在计算机系统中可用，我们必须将其从平滑连续的信号转化为数字流。这个过程是 Edexcel 考试大纲中数据表示主题的核心，也说明了为什么我们需要采样、量化和编码。明白声音是以数值数据的形式被捕获和存储的，你就能直观地理解感知到的音质与磁盘上存储的二进制文件之间的直接联系。
2. Sampling and Sample Rate | 采样与采样率
Sampling is the first step in digitising sound: it measures the amplitude (height) of the analogue wave at fixed intervals. The sample rate, measured in hertz (Hz), tells us how many measurements are taken per second. A higher sample rate captures a truer representation of the original wave because it records more detail. If the sample rate is too low, a phenomenon called aliasing occurs, where high-frequency sounds are misrepresented as lower frequencies, producing distortion. For CD-quality audio, the standard sample rate is 44.1 kHz (44,100 samples per second), but voice recordings for telephony often use just 8 kHz because the human voice has a limited frequency range.
采样是声音数字化的第一步：它每隔固定时间测量模拟波的振幅（高度）。采样率以赫兹（Hz）为单位，表示每秒采集的样本数。更高的采样率能更真实地再现原始波形，因为它记录了更多细节。如果采样率太低，就会发生混叠现象，高频声音被错误地表现为低频，产生失真。CD 品质的音频采用 44.1 kHz（每秒 44,100 次采样）标准采样率，但电话语音录音常只使用 8 kHz，因为人声的频率范围有限。
The Nyquist theorem provides a critical rule: to faithfully reproduce a sound, the sample rate must be at least twice the highest frequency component of the original signal. For example, if you want to capture frequencies up to 20,000 Hz, you need a sample rate of at least 40,000 Hz. This theorem appears regularly in Edexcel-style questions that ask you to justify a chosen sampling frequency or to predict aliasing effects.
奈奎斯特定理给出了一条关键规则：要忠实地重现声音，采样率至少应为原始信号最高频率成分的两倍。例如，如果你想捕获高达 20,000 Hz 的频率，就至少需要 40,000 Hz 的采样率。这个定理经常出现在 Edexcel 风格的考题中，要求你解释所选的采样频率，或预测混叠效应。
3. Bit Depth and Quantisation | 位深度与量化
Once a sample is taken, its amplitude is rounded to the nearest available value in a process called quantisation. The bit depth determines how many discrete amplitude levels are available: an n-bit sample can represent 2ⁿ different levels. Common bit depths include 16-bit (65,536 levels) for CD audio and 24-bit for professional recording. Higher bit depth reduces quantisation error, which is the slight inaccuracy introduced when the true amplitude doesn’t line up exactly with a representable level. This error manifests as background noise, particularly noticeable in quiet passages.
采集到样本后，其振幅会被四舍五入到最接近的可用值，这个过程叫量化。位深度决定了可用的离散振幅级数：n 位样本可表示 2ⁿ 种不同的级别。常见位深度有 CD 音频的 16 位（65,536 级）和专业录音的 24 位。更高的位深度能减小量化误差，即真实振幅无法与可表示级别精确对齐时引入的轻微偏差。这种误差表现为背景噪声，在安静段落尤为明显。
Choosing an appropriate bit depth and sample rate is a balance between quality and file size. Edexcel candidates are expected to explain this trade-off and to calculate the uncompressed storage required for a given piece of audio.
选择合适的位深度和采样率是在音质与文件大小之间寻求平衡。Edexcel 考生需要能够解释这种权衡，并计算一段给定音频所需的未压缩存储空间。
4. Calculating Audio File Size | 计算音频文件大小
Uncompressed audio file size can be calculated using a straightforward formula. The formula ties together sample rate, bit depth, duration, and number of channels:
未压缩音频文件大小可以用一个简单的公式计算。该公式将采样率、位深度、时长和声道数联系在一起：
File size (bits) = Sample rate (Hz) × Bit depth × Duration (s) × Number of channels
文件大小（比特） = 采样率（Hz） × 位深度 × 时长（秒） × 声道数
Remember to convert bits to bytes by dividing by 8, and then to larger units like kilobytes or megabytes as needed. For example, a 3-minute stereo recording at 44.1 kHz with 16-bit depth would produce:
记住，要除以 8 将比特转换为字节，再根据需要转换为千字节或兆字节。例如，一段 3 分钟 44.1 kHz、16 位深度的立体声录音会产生：
Size = 44,100 × 16 × (3 × 60) × 2 = 254,016,000 bits ≈ 30.3 MB
大小 = 44,100 × 16 × (3 × 60) × 2 = 254,016,000 比特 ≈ 30.3 MB
In the exam, you may also be asked to derive any missing variable if the file size and other parameters are given. Practising these conversions will make you fast and confident when tackling structured questions.
在考试中，你也可能需要在给定文件大小和其他参数的情况下推导出缺失的变量。练习这些换算能让你在回答结构化题目时又快又有信心。
5. MIDI – A Different Kind of “Audio” | MIDI – 另一种“音频”
MIDI (Musical Instrument Digital Interface) does not store actual sound waves. Instead, it records a set of instructions for how the music should be played – which notes, how long, how loud, and which instrument. Think of MIDI as a digital score rather than a recording. As a result, MIDI files are extremely small and offer great flexibility in editing and playback, but they cannot reproduce speech or natural sounds, and playback quality depends on the synthesiser hardware.
MIDI（乐器数字接口）并不存储实际的声波，而是记录一组关于如何演奏音乐的命令——哪个音符、时长多长、力度多大、使用哪种乐器。可以把 MIDI 想象成数字乐谱而非录音。因此，MIDI 文件非常小，编辑和回放也具有极大的灵活性，但它无法重现语音或自然声音，而且回放质量取决于合成器硬件。
Edexcel expects you to compare MIDI with sampled digital audio, discussing scenarios where each is appropriate. For recording a live conversation, sampled audio is essential; for composing a synthesised backing track, MIDI is far more storage-efficient.
Edexcel 要求你比较 MIDI 和采样的数字音频，并讨论它们各自适用的场景。对于录制实时对话，采样音频是必不可少的；对于创作合成伴奏轨，MIDI 的存储效率则要高得多。
6. Speech Recognition – When Computers “Listen” | 语音识别 – 当计算机“倾听”
Speech recognition technology converts spoken words into text or executable commands. Although not a core theoretical requirement at AS level, it provides excellent context for questions about input devices, AI, and data processing. The system works by capturing audio through a microphone, digitising it (using the same sampling principles), and analysing acoustic patterns. Modern recognisers use machine learning models to match sound patterns against phonemes – the distinct units of sound in a language – and then reconstruct words using a language model.
语音识别技术将说出的词语转换为文本或可执行命令。虽然这不是 AS 级别的核心理论要求，但它为关于输入设备、人工智能和数据处理的问题提供了极好的背景。该系统的原理是通过麦克风捕获音频，将其数字化（应用相同的采样原理），然后分析声学模式。现代识别器利用机器学习模型将声音模式与音素（语言中的独特声音单位）进行匹配，然后使用语言模型重建单词。
Key issues in speech recognition include background noise, accent variation, and homophones (words that sound the same). These challenges link directly to data representation and the limitations of digital processing, making them fair game for Edexcel’s application-focused questions.
语音识别的关键问题包括背景噪声、口音差异和同音异义词。这些挑战与数据表示及数字处理的局限性直接相关，因此很可能成为 Edexcel 应用题中的素材。
7. Speech Synthesis – When Computers “Speak” | 语音合成 – 当计算机“说话”
On the flip side, speech synthesis produces artificial speech from text. Early systems used concatenative synthesis, stitching together prerecorded fragments of human speech. Today, parametric and neural network-based models generate much more natural intonation. The computation involved mirrors several AS topics: text is first converted to phonetic representations, then rendered as a waveform using digital signal processing. The output must be reconstructed at a suitable sample rate and bit depth for audible playback.
另一方面，语音合成则从文字生成人工语音。早期系统使用拼接合成，将预先录制的人类语音片段拼接起来。如今，基于参数和神经网络模型能生成更自然的语调。这一计算过程涉及多个 AS 主题：文本首先被转换为语音表示，然后使用数字信号处理技术还原为波形。输出必须以合适的采样率和位深度重建，才能被人耳听到。
Understanding synthesis helps reinforce concepts like sampling, DAC (digital-to-analogue conversion), and file formats. Edexcel questions might ask you to explain why a synthesiser requires a DAC, or how a screen reader uses synthetic speech to provide accessibility.
理解语音合成有助于巩固采样、数模转换（DAC）和文件格式等概念。Edexcel 考题可能会让你解释为什么合成器需要 DAC，或屏幕阅读器如何利用合成语音实现无障碍功能。
8. Audio Compression – Making Files Smaller | 音频压缩 – 缩小文件尺寸
Uncompressed audio files, especially high-quality ones, can be enormous. Compression reduces file size for storage and transmission. Lossless compression (such as FLAC) reduces file size without any loss of data by exploiting patterns and statistical redundancy, much like general lossless compression algorithms. When you decompress, the audio is bit-for-bit identical to the original. Lossy compression (like MP3 or AAC) achieves much greater size reductions by discarding audio data that human ears are less likely to notice, based on psychoacoustic models. This removal is irreversible.
未压缩的音频文件，特别是高质量的，可能非常庞大。压缩可以减小文件大小以便存储和传输。无损压缩（如 FLAC）通过利用模式和统计冗余来减小文件大小，且不丢失任何数据，与通用无损压缩算法类似。解压后，音频与原始文件逐比特相同。有损压缩（如 MP3 或 AAC）则基于心理声学模型，丢弃人耳不太可能注意到的音频数据，从而实现更大的压缩率。这种移除是不可逆的。
You must be able to explain the trade-off between file size and perceived quality, and to identify which compression method is suitable for a given purpose – such as streaming music versus archiving original studio recordings. This is a recurring theme in the “consequences of uses of computing” discussions.
你必须能够解释文件大小与感知音质之间的权衡，并能确定哪种压缩方法适合特定目的——比如流媒体传输与原始录音存档。这是“计算机使用的影响”讨论中反复出现的主题。
9. Hardware: From Mic to Speaker | 硬件：从麦克风到扬声器
Speaking and listening in computing are not just about software; they rely on a chain of hardware components. A microphone contains a transducer that converts sound pressure into an analogue voltage signal, which an ADC (analogue-to-digital converter) then samples and digitises. Conversely, a speaker takes a digital signal, passes it through a DAC (digital-to-analogue converter), amplifies the reconstructed analogue wave, and vibrates a diaphragm to create sound. The quality of both ADC and DAC directly affects the fidelity of the reproduced audio.
计算机中的说与听不仅关乎软件，还依赖一系列硬件组件。麦克风包含一个换能器，将声压转换为模拟电压信号，然后由 ADC（模数转换器）进行采样和数字化。相反地，扬声器接收数字信号，通过 DAC（数模转换器），将重建的模拟波形放大，并振动振膜来产生声音。ADC 和 DAC 的质量直接影响重现音频的保真度。
These concepts bridge the “Data representation” and “Hardware” modules in the Edexcel specification, reminding you that abstraction layers work together. When you speak to a voice assistant, the microphone’s sampling process, the CPU’s analysis, and the speaker’s reconstruction all follow the principles you study in Year 12.
这些概念连接了 Edexcel 课程规范中的“数据表示”和“硬件”模块，提醒你抽象层次是协同工作的。当你对语音助手说话时，麦克风的采样过程、CPU 的分析以及扬声器的重建都遵循你在 Year 12 学到的原理。
10. Exam Technique for Audio Questions | 音频类题目的考试技巧
When faced with a question on sound or speech, start by identifying whether it asks about storage calculations, data representation principles, or application analysis. For calculations, write down the formula, substitute the values with units, and double-check your unit conversions. Many marks are lost simply because candidates forget to convert minutes to seconds or bits to bytes. Show your working clearly – Edexcel markers can award method marks even if the final answer is slightly off.
面对关于声音或语音的题目时，首先要判断它是要求存储计算、数据表示原理，还是应用分析。对于计算题，写下公式，代入带单位的数值，并仔细检查单位换算。很多分数只因考生忘记将分钟换算为秒或比特换算为字节而丢失。清晰地展示你的步骤——即便最终答案略有偏差，Edexcel 阅卷人也能给出方法分。
For descriptive questions, use precise terminology: say “sample rate” not “recording speed”, “quantisation noise” not “fuzzy sound”. Link your answer to the theory – Nyquist theorem, psychoacoustic masking, lossy vs lossless – where relevant. If a question presents a scenario, such as a mobile app for voice messaging, discuss technical constraints like minimum sample rate for intelligible speech, suitable compression to save bandwidth, and the role of ADC/DAC. Always keep the target audience or purpose in mind; a podcast service has different requirements than a voice-operated security system.
对于描述性问题，使用精准的术语：说“采样率”而不是“录音速度”，说“量化噪声”而不是“模糊的声音”。在相关时，将答案与理论联系起来——奈奎斯特定理、心理声学掩蔽、有损与无损。如果题目给出一个场景，比如一款语音消息应用，请讨论技术限制，如可懂语音所需的最低采样率、节省带宽的合适压缩方式以及 ADC/DAC 的作用。始终牢记目标受众或用途；播客服务与声控安全系统的需求截然不同。
Finally, practise with past papers and mark schemes where audio crops up – often in data representation and compression contexts. Time yourself on file size calculations so that they become automatic. Your goal is to move confidently through the “speaking and listening” side of computing, showing the examiner that you understand the full journey from real-world sound to binary code.
最后，用往年真题和评分方案进行练习，在数据表示和压缩的情境下尤其如此。自己给文件大小计算计时，使之成为本能。你的目标是自信地应对计算机中“说与听”的部分，向考官展示你理解从现实世界声音到二进制代码的全过程。
Published by TutorHao | Computer Science Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
或截图后用「扫一扫 → 相册」选择本图