The vocoder
From wartime cipher to the robot's singing voice — one invention, two destinies
Related
Every instrument in the previous pages 「made」 sound — this page's effect 「rewrites」 the shape of sound.
「正是这个 dehumanizing(去人性化)的方面,使它在流行音乐与音频娱乐中成为有用的特殊声音效果」.
⭐⭐ This page is the vocoder — first of the effects family. One line: one invention, two destinies.
| Class | Assignment |
|---|---|
| HS class | the effects end of the electrophone system(makes no sound of its own; parasitic on sound sources)· filter banks + envelope followers |
| Origin | ⭐⭐ Homer Dudley(Bell Labs · patent 1938 · Voder demo at the 1939 World's Fair) — born for telecom bandwidth and wartime encryption |
| Two worlds | ⭐⭐ telecom: SIGSALY → LPC at 2.4kbps | music: Moog/Carlos → Kraftwerk's post-human aesthetics |
| Bayin note | Outside the bayin — the first time an effect's 「instrumentality」 was named |
⭐⭐ The main line: one invention, two destinies
「声码器由 Bell Labs 的 Homer Dudley 于 1938 年发明,作为合成人类语音的手段;这项工作发展为通道声码器,用于电信的语音编码以在传输中节省带宽」.
⭐⭐ One circuit, two names in two worlds:
| ⭐ Telecom world | ⭐ Music world | |
|---|---|---|
| Name | voice encoder(secure comms) | vocoder effect(robot voice) |
| Goal | bandwidth · encryption | dehumanisation · futurism |
| Milestones | SIGSALY(1943)→ LPC-10(2.4kbps) | Moog/Carlos → Kraftwerk → Daft Punk |
⭐⭐ Against the site: the Telharmonium's technology 「died」 of its era; the vocoder's 「lives」 in two eras — because it held the needs of two worlds at once.
⭐⭐ The Voder(1939): speech synthesis played by a human
「为了演示其解码器的语音合成能力,Voder(Voice Operation Demonstrator)在 1939–1940 年纽约世博会的 AT&T 馆向公众推出」; 「Voder 由一个电子振荡器(有调音源)与噪声发生器(嘶声)组成、 10 带共鸣滤波器(可变增益放大器)做声道、以及手动控制器(压力感应键控滤波 + 踏板控音高)」.
⭐⭐ 「这是一台操作复杂的机器,但熟练的操作员能产生可辨认的语音」—
⭐⭐ The Voder/Vocoder division: the Voder was 「speech synthesis played by a human」(hand-controlled filters); the Vocoder was 「automatic speech encoding」(machine analysis)— two names, two degrees of automation.
⭐⭐ The principle: discard the pitch, keep the shape
「分析:输入信号经多带滤波器,每带输出由包络跟随器测量;合成:解码器把这些作为控制信号施加到输出滤波通道的相应放大器」; ⭐⭐ 「原始语音信号的瞬时频率信息被丢弃了——对加密而言这并不重要」— ⭐⭐ 「正是这个 dehumanizing(去人性化)的方面,使它在流行音乐中成为有用的特殊声音效果」.
⭐⭐ The sharpest 「engineering-art swap」 on the site: to the encryption engineer, 「discarding pitch」 was the price of compression; to the musician, 「discarding pitch」 is why the machine sings — the insight that sound's 「shape」 can be separated from its 「pitch」 gave birth to both the secure phone and the robot's singing voice.
⭐⭐ The musical turn
「数十年后,Robert Moog 与 Wendy Carlos 造了一台耦合到 Moog 模块化合成器的声码器,把如今标志性的机器人嗓音引入音乐」; 「Carlos 用它为 Stanley Kubrick 的电影 A Clockwork Orange 人声化贝多芬第九交响曲的第四乐章」; 「德国电子先驱 Kraftwerk 在 1974 年专辑 Autobahn 上使用声码器,随后影响了一整代电子音乐家」.
⭐⭐ Against the moog-modular page's Switched-On Bach(Carlos)— Carlos's two 「classical × electronic」 crossings: playing Bach on a synthesizer(1968), singing Beethoven through a vocoder(A Clockwork Orange)— each rewrote what electronic instruments could do.
⭐⭐ Three ways a machine sings: vocoder / Auto-Tune / talkbox
| ⭐ Vocoder | Auto-Tune | ⭐⭐ Talkbox | |
|---|---|---|---|
| Principle | spectral transplant(two inputs) | pitch correction(one input) | acoustic filtering(physical) |
| Pitch | untouched | changed | changed(by lips) |
| Where filtered | electronic filter banks | software | the performer's mouth |
| Landmarks | Kraftwerk · Daft Punk | Cher《Believe》· T-Pain | Frampton · Zapp |
⭐⭐ The talkbox's 「filtering happens inside the performer's mouth」 — against the site's 「body as instrument」 lineage(jaw harp/throat carvings/kalimba resonators)— the talkbox is the most 「bodily」 effect of the electronic age.
⭐⭐ The telecom continuation: parameters, not waveforms
「早期数字电话系统使用脉冲编码调制,需要 64 kbps 传输一路语音—— 那是很大的带宽」; 「基于线性预测编码(LPC)的声码器通过建模声道、只传输重建语音所需的参数而非完整音频波形,大幅降低带宽——美国政府标准 LPC-10 只用 2.4 kbps」.
⭐⭐ 「只传参数不传波形」 — the vocoder's philosophy became the foundation of digital communication: against the Prophet-5's patch memory(「remember parameters, not audio」)— 「parameterisation」 is the fundamental idea of digital audio, and the vocoder was its first massive application.
⭐⭐ Cher 1998 and the software version of flaw-into-signature
「重度应用把人声锁到最近的半音太快,把歌手的音高规训到不自然的机器完美—— 常被误认为声码器效果,合成的 Auto-Tune 摆音由 Cher 的 1998 年金曲 Believe 推广,此后被 Kanye West · T-Pain · Will.i.am 使用(可以说过度使用)」.
⭐⭐ Against the TB-303(flaw into signature) — Auto-Tune's 「over-correction」 is also a flaw become signature — but a 「software flaw」, not a 「hardware flaw」. Twice the 「wrong use」 was adopted by music: the 303 took 7 years, Believe took 1.
Recap: three sentences on the vocoder
| Q | A |
|---|---|
| ⭐ The key point | ⭐⭐ one invention, two destinies — telecom(SIGSALY → LPC at 2.4kbps)and music(Moog/Carlos → Kraftwerk); ⭐⭐ 「discarding pitch, keeping shape」 — the dehumanising IS the musical effect |
| ⭐⭐ The Voder | ⭐⭐ 1939 World's Fair: speech synthesis played by a human(pressure-keyed filters + pitch pedal)— Voder(manual)vs Vocoder(automatic) |
| ⭐⭐ The principle | multiband filters + envelope followers → control signals onto the carrier's filter channels; ⭐⭐ 「instantaneous frequency information was discarded」 — the musical effect comes from the engineering's discard |
| ⭐⭐ The turn | Moog/Carlos(Beethoven in A Clockwork Orange)→ Kraftwerk Autobahn(1974) — the robot voice enters music |
| ⭐⭐ The three-way split | Vocoder spectral-transplant|Auto-Tune pitch-correction|Talkbox mouth-filtering — ⭐⭐ the talkbox filters inside the performer's mouth(the most 「bodily」 effect) |
| ⭐⭐ Telecom continuation | LPC-10 at 2.4kbps(vs PCM 64kbps) — ⭐⭐ 「parameters, not waveforms」= the fundamental idea of digital audio(against the Prophet-5's patch memory) |
| ⭐⭐ Cher 1998 | Auto-Tune over-correction as signature — ⭐⭐ the software version of flaw-into-signature(303 took 7 years, Believe took 1) |
| ⭐ Daft Punk | the 「singing robot」 image — often performed on a talkbox — ⭐ what you see and hear is not what it is |
| HS | the effects end of the electrophone system(parasitic on sound sources) |
The vocoder's story is 「one circuit, two reincarnations」: first born as the secure phone, reborn as the robot's singing voice. ⭐⭐ Its philosophy — 「parameters, not waveforms」 — still governs digital audio: you think you are listening to sound; often you are listening to parameters.
声码器属电鸣系统的效果端(不独立发声——寄生在音源上)。原理:调制信号(人声)经多带滤波(8–20 频带)+ 包络跟随器 → 控制信号施加到载波(合成器波形/噪声)的对应滤波通道——输出音域等于载波音源的音域(典型 C1–C5)。声码器不修正音高——它丢弃调制信号的瞬时频率、只保留频谱形状。Voder(1939 世博)为手动版本;SIGSALY(1943)为战时加密版本;LPC-10(2.4kbps)为现代数字电话地基。
Listen in the library
All tracks from midicn-lib — inline badge is the license tier (C1 commercial / C2 non-commercial / C3 study only).
Sources
- ⭐⭐ 主线:一个发明,两种命运(本页最该记住的一条) —— 「声码器由 Bell Labs 的 Homer Dudley 于 1938 年发明,作为合成人类语音的手段」;「这项工作发展为通道声码器,用于电信的语音编码以在传输中节省带宽」;⭐⭐ 两条命运:电信世界(SIGSALY 战时加密 → LPC 语音压缩成为数字电话标准)与音乐世界(Moog/Carlos → Kraftwerk 的后人类美学)——同一个电路,两个世界的两种「名」
- ⭐⭐ Voder(1939 世界博览会)与 Vocoder 的分工 —— 「为了演示其解码器的语音合成能力,Voder(Voice Operation Demonstrator)在 1939–1940 年纽约世博会的 AT&T 馆向公众推出」;「Voder 由一个电子振荡器(有调音源)与噪声发生器(嘶声)组成、10 带共鸣滤波器(可变增益放大器)做声道、以及手动控制器(压力感应键控滤波 + 踏板控音高)」;⭐⭐ 「这是一台操作复杂的机器,但熟练的操作员能产生可辨认的语音」——Voder 是「人来演的语音合成」,Vocoder 是「自动的语音编码」——两个名字,两种自动化程度
- ⭐⭐ 原理(两个阶段) —— 「分析:输入信号经多带滤波器,每带输出由包络跟随器测量,包络跟随器的信号被传送到解码器」;「合成:解码器把这些作为控制信号施加到输出滤波通道的相应放大器」;⭐⭐ 「原始语音信号的瞬时频率信息被丢弃了——对声码器作为加密辅助的原始用途而言这并不重要」——⭐⭐ 「正是这个 dehumanizing(去人性化)的方面,使它在流行音乐与音频娱乐中成为有用的特殊声音效果——丢弃「音高」保住「形状」:音乐效果正来自工程的丢弃
- ⭐⭐ 音乐化的转折(Moog/Carlos/Kraftwerk) —— 「数十年后,Robert Moog 与 Wendy Carlos 造了一台耦合到 Moog 模块化合成器的声码器,把如今标志性的机器人嗓音引入音乐」;「Carlos 用它为 Stanley Kubrick 的电影 A Clockwork Orange 人声化贝多芬第九交响曲的第四乐章」;⭐⭐ 「德国电子先驱 Kraftwerk 在 1974 年专辑 Autobahn 上使用声码器,随后影响了一整代电子音乐家——Herbie Hancock · Afrika Bambaataa……」
- ⭐⭐ vocoder / Auto-Tune / talkbox 三分(一页说清三种「机器唱歌」) —— 「声码器把一个信号的频谱特征施加到另一个信号——它不改音高,从两个输入创造新声音」;「Auto-Tune 是音高修正工具——分析演唱并把跑调的音符移到最近的正确音高——重度使用+快速修正产生 T-Pain 式的平滑阶梯效果」;⭐⭐ 「talkbox 是物理装置:把吉他音箱或合成器的声音经管子送进演奏者嘴里,演奏者用嘴唇舌头下颚塑造声音,麦克风拾取结果——滤波发生在演奏者口腔内而非电子滤波器组」(Peter Frampton 的 Show Me the Way · Zapp 的目录是 talkbox 地标)
- ⭐⭐ 电信世界的延续(LPC) —— 「早期数字电话系统使用脉冲编码调制,需要 64 kbps 传输一路语音——那是很大的带宽」;「基于线性预测编码(LPC)的声码器通过建模声道、只传输重建语音所需的参数而非完整音频波形,大幅降低带宽——美国政府标准 LPC-10 只用 2.4 kbps」——⭐⭐ 「只传参数不传波形」——声码器的哲学成了整个数字通信的地基
- ⭐⭐ CHER 1998 与 Auto-Tune 的滥用 —— 「重度应用把人声锁到最近的半音太快,把歌手的音高规训到不自然的机器完美——常被误认为声码器效果,合成的 Auto-Tune 摆音由 Cher 的 1998 年金曲 Believe 推广,此后被 Kanye West · T-Pain · Will.i.am 使用(可以说过度使用)」——⭐⭐ 对照 TB-303(缺陷成签名):「Auto-Tune 的「过度修正」也是缺陷成签名——但它是「软件缺陷」而非「硬件缺陷」
- Daft Punk —— 「Daft Punk 用声码器人格化他们的『唱歌机器人』形象,在音乐上融合人声与合成器——不过法国组合常用 talkbox——经管子说话的物理盒子来创造类声码器效果」——⭐ 连「机器人形象」本身都可以是 talkbox 演的:眼见耳听都未必为实
- 「声码器属电鸣系统的效果端(不独立发声——寄生在音源上)」依 Hornbostel–Sachs 分类
Used for fact-checking only; all prose is original.