📚 Oral/Aural Exam Preparation in Computing | 计算机口语/听力备考专项
In IGCSE Computer Science, oral and aural topics centre on how computers handle spoken input and produce sound output. This revision guide covers speech recognition, audio sampling, digital sound representation, text-to-speech, storage requirements, and typical exam-style questions that test your understanding of how computers ‘speak’ and ‘listen’. Prepare for both theory and practical-based questions with clear bilingual explanations.
在IGCSE计算机科学中，口语和听力专题聚焦于计算机如何处理语音输入和产生音频输出。这份备考指南涵盖语音识别、音频采样、数字声音表示、文本转语音、存储需求以及典型的考试题型，帮助你理解计算机如何“说话”和“倾听”。中英双语讲解，助你准备好理论和实践类问题。
1. How a Computer ‘Hears’: Speech Recognition Basics | 计算机如何“听”：语音识别基础
Speech recognition is the process by which a computer converts spoken words into digital text or commands. A microphone captures analogue sound waves; an analogue-to-digital converter (ADC) samples the signal; the sampled data is analysed against stored phoneme patterns and language models to identify words. Accuracy depends on sampling rate, bit depth, and noise reduction algorithms. Voice assistants and dictation software rely on continuous speech recognition, which is more complex than isolated word recognition.
语音识别是计算机将口语转换为数字文本或指令的过程。麦克风捕捉模拟声波；模数转换器(ADC)对信号采样；采样数据与存储的音素模式和语言模型进行比对识别词语。准确率取决于采样率、位深和降噪算法。语音助手和听写软件依赖连续语音识别，这比孤立词语识别更复杂。
2. Analogue to Digital: Sampling and Quantisation | 模数转换：采样与量化
Sound is analogue by nature—a continuous wave. To store it digitally, the wave is sampled at regular intervals and each sample is quantised to a discrete amplitude value. The sampling rate, measured in hertz (Hz), is the number of samples per second. Common rates: 44.1 kHz for CD quality, 8 kHz for telephone. Quantisation divides the amplitude range into a finite number of levels, determined by bit depth (e.g., 16-bit audio gives 65,536 levels). Higher sampling rates and bit depths mean better quality but larger file sizes.
声音本质上是模拟的——连续的波形。要数字化存储，需定期采样并于每一采样点量化成离散振幅值。采样率以赫兹(Hz)为单位，即每秒采样数。常见速率：CD品质为44.1 kHz，电话为8 kHz。量化将振幅范围划分成有限级数，由位深决定（如16位音频提供65,536个等级）。采样率和位深越高，质量越好，但文件体积越大。
3. Audio File Size and Bit Rate Calculation | 音频文件大小与比特率计算
Exam questions often ask you to calculate uncompressed audio file size. The formula is:
File size (bits) = sampling rate (Hz) × bit depth × number of channels × duration (seconds)
For example, a 1-minute stereo track at 44.1 kHz and 16-bit: size = 44,100 × 16 × 2 × 60 = 84,672,000 bits, which is about 10.09 MB (divide by 8 for bytes, then by 1,048,576 for MB). Bit rate is simply the bits transferred per second: sample rate × bit depth × channels. For streaming or spoken audio, lower bit rates are often used to conserve bandwidth without sacrificing much intelligibility.
试题常要求计算未压缩音频文件大小。公式为：
文件大小(位) = 采样率(Hz) × 位深 × 声道数 × 时长(秒)
例如，一段1分钟立体声，采样率44.1 kHz、16位：大小 = 44,100 × 16 × 2 × 60 = 84,672,000位，约10.09 MB（除以8得字节，再除以1,048,576得MB）。比特率即每秒传输的位数：采样率 × 位深 × 声道数。流媒体或语音音频常采用较低比特率，以节省带宽而不显著影响可懂度。
4. Speech Synthesis: From Text to Spoken Word | 语音合成：从文字到语音
Text-to-speech (TTS) systems convert written text into artificial speech. A typical pipeline involves text normalisation (expanding abbreviations, numbers), linguistic analysis (determining prosody and phonemes), and waveform generation. Two methods are common: concatenative synthesis, which stitches together pre-recorded speech segments, and formant synthesis, which generates sound from scratch using acoustic models. Modern systems use deep neural networks to produce highly natural voices. TTS is used in screen readers, navigation devices, and virtual assistants.
文本转语音(TTS)系统将书面文字转换为人工语音。典型流程包括文本规范化（扩展缩写、数字）、语言学分析（确定韵律和音素）以及波形生成。常见方法有两种：拼接合成，将预录语音片段拼接；共振峰合成，利用声学模型从头生成声音。现代系统使用深度神经网络产生非常自然的语音。TTS应用于屏幕阅读器、导航设备和虚拟助手。
5. Human Speech vs. Computer Speech: Features & Limitations | 人声与计算机语音：特征与局限
Human speech carries emotion, intonation, and nuances that are difficult for machines to replicate perfectly. Computers can now mimic prosody and stress patterns, but may still sound slightly unnatural. Speech recognition struggles with accents, background noise, homophones (e.g., ‘write’ and ‘right’), and overlapping speech. Conversely, TTS may mispronounce homographs (e.g., ‘lead’ as metal vs. to guide). Understanding these limitations is key for both design and exam analysis questions.
人声带有情感、语调和细微差别，机器难以完美复现。计算机如今可模仿韵律和重音模式，但仍可能略显不自然。语音识别在口音、背景噪音、同音异义词（如“write”和“right”）以及重叠说话时表现欠佳。反之，TTS可能读错同形异义词（如“lead”作金属或引导）。理解这些局限对设计和考试分析题至关重要。
6. Audio Encoding and Compression | 音频编码与压缩
Raw audio files are huge. Compression reduces size through lossless or lossy techniques. Lossless formats (e.g., FLAC, ALAC) preserve all original data by removing statistical redundancy. Lossy compression (e.g., MP3, AAC) discards sounds the human ear is less sensitive to, based on psychoacoustic models. Speech-specific codecs like AMR or Opus achieve very low bit rates while keeping speech intelligible, essential for VoIP and mobile telephony. Know how compression affects quality and file size; exams may ask you to compare formats.
原始音频文件体积庞大。压缩通过无损或有损技术减小尺寸。无损格式（如FLAC、ALAC）通过去除统计冗余保留所有原始数据。有损压缩（如MP3、AAC）基于心理声学模型丢弃人耳不太敏感的声音。语音专用编解码器如AMR或Opus在极低比特率下仍保持语音可懂度，对VoIP和移动电话至关重要。需了解压缩如何影响质量与文件大小；考试可能要求比较格式。
7. Input Devices for Oral Tasks | 口语任务的输入设备
Microphones are the primary input device for oral interaction. Types include dynamic, condenser, and MEMS microphones in smartphones. Key specifications: frequency response (range of frequencies captured), sensitivity, and directionality (omnidirectional, cardioid). Noise-cancelling microphones use multiple elements to suppress background sound. For oral exams in computing contexts, you must explain how the quality of the microphone impacts speech recognition accuracy and how signal processing filters noise.
麦克风是口语交互的主要输入设备。类型包括动圈式、电容式和智能手机中的MEMS麦克风。关键规格：频率响应（捕获的频率范围）、灵敏度和方向性（全向、心形）。降噪麦克风利用多元件抑制背景声。计算机学科口语化考试中，你需要解释麦克风质量如何影响语音识别准确度，以及信号处理如何滤除噪声。
8. Output Devices for Aural Feedback | 听觉反馈的输出设备
Speakers and headphones convert digital audio signals back into analogue sound waves via a digital-to-analogue converter (DAC) and an amplifier. In aural exam scenarios, the focus is on clarity and latency. Latency is the delay between audio generation and its playback; high latency can disrupt real-time communication. Digital speakers may include built-in DACs. Understanding the signal path helps in troubleshooting audio issues and in explaining how computers ‘speak’ to users through sound cards and drivers.
扬声器和耳机通过数模转换器(DAC)与放大器将数字音频信号转换回模拟声波。在听力类考试场景中，重点是清晰度和延迟。延迟是指音频生成与播放之间的时间差；高延迟会干扰实时交流。数字音箱可能内置DAC。理解信号路径有助于解决音频问题，并解释计算机如何通过声卡和驱动程序向用户“说话”。
9. Voice User Interfaces (VUI) and Natural Language Processing | 语音用户界面与自然语言处理
Voice user interfaces allow hands-free interaction. They combine automatic speech recognition (ASR) with natural language processing (NLP). NLP interprets the meaning behind spoken words, handling intent recognition, entity extraction, and dialogue management. For example, “What’s the weather like tomorrow?” triggers a weather API. CIE exam questions may ask you to describe how a smart speaker works or to evaluate the usability challenges of VUIs, such as misinterpretation, privacy concerns, and lack of visual feedback.
语音用户界面支持免提交互。它们结合自动语音识别(ASR)与自然语言处理(NLP)。NLP解读口语背后的含义，处理意图识别、实体提取和对话管理。例如，“明天天气如何？”触发天气API。CIE试题可能要求你描述智能音箱的工作原理，或评估VUI的可用性挑战，如误识别、隐私担忧和缺乏视觉反馈。
10. Exam-Style Question: Audio Storage Calculation | 考试题型：音频存储计算
Question: A dictation app records mono audio at 16 kHz sampling rate and 8-bit depth. A student records a 30-second spoken answer. Calculate the file size in kilobytes. Show your steps.
Answer: Size in bits = 16,000 × 8 × 1 × 30 = 3,840,000 bits. Convert to bytes: 3,840,000 / 8 = 480,000 bytes. Convert to KB: 480,000 / 1,024 ≈ 468.75 KB. Always check units and whether the question expects KB or KiB. Practice similar calculations with stereo, different durations, and varying bit depths to reinforce the formula.
题目：某听写应用以16 kHz采样率、8位位深录制单声道音频。一名学生录制了30秒口语作答。计算文件大小，以千字节(KB)表示。写出步骤。
解答：大小(位) = 16,000 × 8 × 1 × 30 = 3,840,000位。转换为字节：3,840,000 / 8 = 480,000字节。转换为KB：480,000 / 1,024 ≈ 468.75 KB。务必核对单位，注意题目要求KB还是KiB。通过立体声、不同时长和位深的类似计算练习，巩固公式。
11. Common Misconceptions and Pitfalls | 常见误区与陷阱
Students often confuse sampling rate with bit rate, or think higher sampling always doubles the file size regardless of bit depth. Another mistake is forgetting to account for multiple channels in size calculations. When discussing speech recognition, they might assume the computer ‘understands’ like a human, neglecting the algorithmic and probabilistic nature. In TTS, overlooking prosody generation leads to overestimation of naturalness. Exam answers should reflect precise technical language and awareness of these subtle but crucial distinctions.
学生常混淆采样率与比特率，或误以为提高采样率总会使文件大小翻倍而不考虑位深。另一个错误是在计算大小时忘记乘声道数。讨论语音识别时，可能假定计算机像人一样“理解”，忽略了算法和概率本质。在TTS中，忽视韵律生成会高估自然度。答卷应体现准确的技术用语，并注意这些细微而关键的区别。
12. Quick Revision Summary for Oral/Aural Topics | 口语/听力专题快速复习摘要
| Concept 概念 | Key Points 要点 |
|---|---|
| Sampling 采样 | Rate (Hz) × bit depth × channels × time = file size |
| Speech Recognition 语音识别 | ADC → pattern matching → language model; affected by noise |
| TTS 文本转语音 | Text normalisation → phoneme generation → waveform synthesis |
| Compression 压缩 | Lossless vs lossy; speech codecs keep low bit rate for voice |
| VUI 语音用户界面 | ASR + NLP; hands-free but privacy and error challenges |
This table condenses the main themes you must master for aural/oral computing questions. Revise each row and be ready to apply formulas, explain processes, and evaluate advantages and disadvantages.
此表浓缩了听力/口语计算机问题必须掌握的主要主题。复习每行内容，准备应用公式、解释流程并评价优缺点。
Published by TutorHao | Computing Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
屏轩国际教育cambridge primary/secondary checkpoint, cat4, ukiset,ukcat,igcse,alevel,PAT,STEP,MAT, ibdp,ap,ssat,sat,sat2课程辅导，国外大学本科硕士研究生博士课程论文辅导