Bell Labs Debuts Voder Synthesizer, Rules Out Commercial Product

The Voder: 1939, the worlds first electronic voice synthesizer — Brian Roemmele

Brian Roemmele

Bell Telephone Laboratories developed the intricate voice synthesizer as an educational showcase of underlying speech compression research rather than a standalone commercial product.

AI-assisted summary of The Voder: 1939, the worlds first electronic voice synthesizer — original source. Check the original for full context.

Source published:

Key ideas

  • Developed by Bell Telephone Laboratories for the 1939 World's Fair, the Voder generated synthetic human speech electronically using two energy sources and ten filter circuits to produce twenty acoustic components. (0:18)
  • Operating the machine demanded extraordinary motor coordination; articulating a single word like 'concentration' required forming 13 successive sounds, five wrist bar strokes, and up to five foot pedal adjustments. (3:36)
  • Training was exceptionally rigorous, requiring a full year of daily practice, with only 28 women qualifying as certified operators out of approximately 320 auditioned candidates. (3:36)
  • Bell Labs did not intend to commercialize the Voder directly, positioning it as an educational showcase derived from broader telecommunications research into speech compression and transmission. (4:40)

Detailed notes

## Background Published on YouTube by Brian Roemmele on November 21, 2016, this archival recording captures a live radio broadcast demonstrating the Voder, the world's first electronic voice synthesizer developed in 1939 by Homer Dudley at Bell Telephone Laboratories. According to the description, the manually operated system was showcased at the 1939 New York World's Fair and required extensive physical coordination across finger keys, pedals, and levers to synthesize continuous human speech.

## TL;DR In this live radio demonstration, host Dr. Caldwell, Bell Labs representative Mr. Garrett, and expert operator Miss Helen Harper showcase the acoustic capabilities and underlying architecture of the Voder. Miss Harper demonstrates how the instrument manipulates pitch, emotional inflection, and vibrato to recite verses, laugh, and sing 'Auld Lang Syne,' before demonstrating its ability to speak French and mimic barnyard animals. The broadcast also highlights the grueling human skill needed to operate the synthesizer—requiring simultaneous coordination of fingers, wrist bars, and foot pedals mastered by only 28 of 320 auditioned operators—while explaining that the machine serves as an educational byproduct of telecommunications research into speech compression.

## Atmosphere & style The broadcast exudes the crisp, polite elegance of golden-age radio, balancing educational scientific inquiry with jaunty, lighthearted banter among the hosts. Listening to it creates a charmingly uncanny experience as the refined mid-century presenters interact naturally with the raspy, groaning, and occasionally eerie acoustic utterances of the machine. The mood oscillates between whimsical amusement during the Voder's dismal nursery rhymes and animal imitations, and genuine awe at the virtuosic physical dexterity required to operate the synthetic voice.

## Bulletin - [00:18](https://www.youtube.com/watch?v=TsdOej_nC1M&t=18s) Developed by Bell Telephone Laboratories for the 1939 World's Fair, the Voder generated synthetic human speech electronically using two energy sources and ten filter circuits to produce twenty acoustic components. - [03:36](https://www.youtube.com/watch?v=TsdOej_nC1M&t=216s) Operating the machine demanded extraordinary motor coordination; articulating a single word like 'concentration' required forming 13 successive sounds, five wrist bar strokes, and up to five foot pedal adjustments. - [03:36](https://www.youtube.com/watch?v=TsdOej_nC1M&t=216s) Training was exceptionally rigorous, requiring a full year of daily practice, with only 28 women qualifying as certified operators out of approximately 320 auditioned candidates. - [04:40](https://www.youtube.com/watch?v=TsdOej_nC1M&t=280s) Bell Labs did not intend to commercialize the Voder directly, positioning it as an educational showcase derived from broader telecommunications research into speech compression and transmission.

## Key takeaways - [00:18](https://www.youtube.com/watch?v=TsdOej_nC1M&t=18s) The Voder synthesizes speech electronically through two energy sources and a bank of ten filter circuits producing twenty acoustic components. - [00:18](https://www.youtube.com/watch?v=TsdOej_nC1M&t=18s) Operators could dynamically alter sentence stress and inflection to express nuanced contextual meanings, such as shifting emphasis across the words in 'She saw me.' - [00:18](https://www.youtube.com/watch?v=TsdOej_nC1M&t=18s) The synthesizer could readily adjust pitch registers, transitioning across standard, high-pitched, and resonant bass ranges. - [00:18](https://www.youtube.com/watch?v=TsdOej_nC1M&t=18s) Applying vibrato to sustained vowel outputs softens the raw, robotic buzz into a musical singing tone capable of performing songs like 'Auld Lang Syne.' - [03:36](https://www.youtube.com/watch?v=TsdOej_nC1M&t=216s) Synthesizing speech required intricate physical coordination across multiple limbs, using finger keys, foot pedals, and a wrist bar with split-second timing. - [03:36](https://www.youtube.com/watch?v=TsdOej_nC1M&t=216s) Mastering the Voder's control interface required a full year of continuous daily practice. - [03:36](https://www.youtube.com/watch?v=TsdOej_nC1M&t=216s) Out of roughly 320 candidates auditioned by Bell Telephone Laboratories, only 28 demonstrated the specialized aptitude needed to become certified operators. - [04:40](https://www.youtube.com/watch?v=TsdOej_nC1M&t=280s) Because the device synthesizes fundamental speech sounds rather than recorded words, it is capable of speaking foreign languages such as French. - [04:40](https://www.youtube.com/watch?v=TsdOej_nC1M&t=280s) The acoustic range of the system extended beyond speech to animal mimicry, including cow lowing and pig grunting. - [04:40](https://www.youtube.com/watch?v=TsdOej_nC1M&t=280s) The Voder was developed as a public exhibition piece for the 1939 New York World's Fair and the Golden Gate International Exposition rather than a commercial device. - [04:40](https://www.youtube.com/watch?v=TsdOej_nC1M&t=280s) The synthesizer served as a showcase for Bell Labs' foundational telecommunications research into speech compression, acoustic analysis, and electrical voice transmission.

## Detailed walkthrough ### [00:18](https://www.youtube.com/watch?v=TsdOej_nC1M&t=18s) Inflection, Voice Range, and the Singing Tone

The demonstration opens with an explanation of the core technical architecture of the Voder: ten filter circuits combined with two energy sources produce a total of twenty separate acoustic components utilized to assemble human speech sounds. To show how these building blocks operate in practice, Mr. Garrett and operator Miss Helen Harper demonstrate how individual words assemble into complete sentences. Helen commands the machine to utter the phrase, "She saw me." When the initial rendition sounds flat and robotic, Mr. Garrett requests varying shades of human expression by posing targeted questions. Helen subtly manipulates the controls to shift the sentence stress to fit the context: emphasizing the subject for *"Who saw you?"* (“*She* saw me”), shifting stress to the object for *"Whom did she see?"* (“She saw *me*”), and highlighting the verb when asked, *"Did she see you or hear you?"* (“She *saw* me”).

```mermaid flowchart LR accTitle: Voder Acoustic Architecture accDescr: Diagram showing how two energy sources and ten filter circuits combine into speech components. subgraph Sources["Energy Sources"] E1["Hiss / Noise Source"] E2["Buzz / Voice Pitch"] end subgraph Filtering["Filter Bank"] F["10 Filter Circuits"] end subgraph Output["Acoustic Output"] C["20 Speech Components"] S["Continuous Synthetic Speech"] end Sources --> F F --> C C --> S ```

The presenters show that the machine is not limited to a single register. By making a straightforward adjustment to the controls, Miss Harper switches the voice quality. The Voder delivers the greeting, "Greetings, everybody," first in a standard voice, then in a high pitch, and finally in a resonant bass register. Mr. Garrett jokes about the vocal awkwardness of adolescent boys:

> “When a boy's voice changes he's never quite sure whether it's going to be a tenor or a bass, and the Voder, being still a comparatively young man, also has his moments of uncertainty.”

The Voder is then asked to recite "Mary Had a Little Lamb," but recites it with such dreary, pitch-shifting inflections that Mr. Garrett remarks, "Gosh, Voder, you sounded awfully dismal! Snap out of it, let's hear you laugh!" The resulting synthetic laughter sounds so creaky and decrepit that Garrett playfully wonders whether he was mistaken about its youth, asking, "Do you really feel as old as you sound?" to which the synthesizer drolly groans, "Yes, I do."

Turning to more melodious capabilities, Garrett introduces vibrato into the Voder's sustained vowel output. Without pitch modulation, an extended vowel sounds stark, robotic, and unnatural, but adding vibrato instantly transforms the sustained buzz into a recognizable singing tone. Demonstrating this musicality, the Voder tests its vocal scale and performs a rendition of "Auld Lang Syne," prompting the narrator to remark that the machine sings the classic tune better than he ever could.

[03:36](https://www.youtube.com/watch?v=TsdOej_nC1M&t=216s) Operating Demands and the Rigors of Voder Training

The broadcast host, Dr. Caldwell, asks Miss Harper to explain the physical complexity required to produce a single synthesized word. Miss Harper reveals that operating the Voder demands immense motor coordination and concentration across multiple limbs:

> “Well, for example, in producing the word 'concentration' on the Voder, I have to form 13 different sounds in succession, make five up and down movements of the wrist bar, and vary the position of the foot pedals from three to five times, according to what expression I want the Voder to give the word.”

All of these intricate fingerings, wrist movements, and pedal adjustments must be executed with exact timing. Miss Harper explains that mastering the instrument required a full year of constant daily practice.

- Bell Telephone Laboratories auditioned roughly 320 candidates to find operators for the machine. - Out of those 320 women, only 28 attained the level of skill required to become certified expert operators. - As Mr. Garrett notes regarding the low qualification rate: “I can answer that, Dr. Caldwell, by saying that in operating the Voder, a girl needs a peculiar combination of particular talents which is not too common.”

[04:40](https://www.youtube.com/watch?v=TsdOej_nC1M&t=280s) Foreign Speech, Animal Mimicry, and Practical Applications

Dr. Caldwell asks whether the machine is limited to English or capable of foreign tongues. Mr. Garrett explains that the system can mimic any language spoken by the person at the controls, which Miss Harper demonstrates by having the Voder pronounce the French phrase *"Parlez-vous français?"* with clear articulation. Mr. Garrett asserts:

> “You see, Doctor, the Voder can do practically anything that can be done with a human voice.”

The demonstration tests the acoustic limits of the instrument with novel sounds: - Imitating barnyard animals, including the recognizable lowing of a cow and the grunting squeal of a pig. - Demonstrating the Voder's basic "education" by having it systematically recite the letters of the alphabet from *A* through *F*.

When Dr. Caldwell inquires about the eventual commercial applications of the machine, Mr. Garrett clarifies that Bell Telephone Laboratories does not foresee direct commercialization for the instrument itself. Instead, the Voder was created primarily as an educational showcase for the 1939 New York World's Fair and the Golden Gate International Exposition in San Francisco. Garrett explains that the synthesizer exists as a public-facing byproduct of Bell Labs' broader ongoing telecommunications research into speech compression, electrical voice transmission, and acoustic analysis.

The demonstration concludes with Miss Harper guiding the Voder to bid the radio audience farewell, delivering the sign-off: "Good evening, radio audience."

## Notable quotes > [04:40](https://www.youtube.com/watch?v=TsdOej_nC1M&t=280s) The Voder can do practically anything that can be done with a human voice. — Mr. Garrett

> [03:36](https://www.youtube.com/watch?v=TsdOej_nC1M&t=216s) In operating the Voder, a girl needs a peculiar combination of particular talents which is not too common. — Mr. Garrett

## Mentioned The Voder, Homer Dudley, Bell Telephone Laboratories, 1939 New York World's Fair, Golden Gate International Exposition, Dr. Caldwell, Mr. Garrett, Miss Helen Harper, Auld Lang Syne, Mary Had a Little Lamb

## Video description The Voder was the worlds first voice electronic synthesizer. In 1939 Homer Dudley working at Bell Telephone Laboratories in Murray Hill, New Jersey began to publicly demonstrate the Voder, the first electronic device that could generate continuous human speech electronically.

The Voder was designed for the 1939 World Fair in New York as a showcase of the advanced work being done at Bell Laboratories. It was a manually operated system requiring training, ten finger, two foot paddles, a knee leaver and arm switch to generate sounds. The sounds quality actually was better than most voice synthesizers all the way up to the late 1990s.

This is audio from a radio program recorded live during the broadcast.

[Source](https://www.youtube.com/watch?v=TsdOej_nC1M)

中文简报

贝尔实验室推出Voder合成器，排除商业化可能 贝尔电话实验室开发了这款复杂的语音合成器，将其作为底层语音压缩研究的科普展示，而非独立的商业产品。 在这场电台现场演示中，主持人考德威尔博士、贝尔实验室代表加勒特先生以及专业操作员海伦·哈珀小姐展示了Voder的声音性能和底层架构。哈珀小姐演示了该设备如何调控音高、情感语调和颤音来朗诵诗句、发出笑声并演唱《友谊地久天长》，随后还展示了其讲法语以及模仿农场动物叫声的能力。广播还重点指出了操控该合成器所需的极高人体技能——需要手指、腕托和脚踏板的同时协调，在试音的320名操作员中仅有28人完全掌握——同时解释了该机器只是电信语音压缩研究的一项科普副产物。 Voder由贝尔电话实验室为1939年世博会开发，利用两个能源和十个滤波电路生成二十种声音成分，以电子方式产生合成的人类语音。 (0:18) 操作该机器需要非凡的运动协调能力；仅清晰说出像“concentration”这样的单个单词，就需要形成13个连续的声音、按压5次腕托并进行多达5次脚踏板调节。 (3:36) 培训极其严苛，需要一整年的每日练习，在约320名参与选拔的候选人中，最终仅有28名女性合格成为认证操作员。 (3:36) 贝尔实验室并未打算直接将Voder商业化，而是将其定位为源自语音压缩与传输这一更广泛电信研究领域的科普成果展示。 (4:40)

Reading summary

In a 1939 demonstration presented by Bell Telephone Laboratories, the Voder demonstrated the feasibility of artificial voice synthesis by assembling complex human speech through electrical filtering, manual control, and acoustic modulation. Acoustic Architecture and Vocal Expressiveness (0:18) The Voder generated speech components through a core architecture combining two energy sources—a hiss noise source and a buzz voice pitch—with ten filter circuits. These inputs yielded twenty distinct acoustic components that an operator could assemble into continuous synthetic speech. Through targeted adjustments, the machine could vary pitch across registers from high tones to deep bass, modulate sentence stress based on conversational context, and apply vibrato to sustained vowels to turn flat robotic sounds into singing tones, as demonstrated in a performance of 'Auld Lang Syne.' Operational Complexity and Specialized Operator Training (3:36) Operating the synthesizer required high motor coordination and continuous physical control across the hands, wrists, and feet. Operator Miss Helen Harper noted that voicing a single word such as 'concentration' entailed forming thirteen consecutive sounds, executing five up-and-down wrist bar shifts, and altering foot pedal positions three to five times depending on the desired inflection. Bell Telephone Laboratories auditioned roughly 320 candidates for the role, yet only 28 attained certification as expert operators following a full year of intensive daily practice. Versatility and Bell Labs' Research Purpose (4:40) Beyond English sentences, the synthesizer could mimic barnyard animal noises like a cow's lowing or a pig's squeal, recite the alphabet, and reproduce foreign languages such as French when guided by the operator. Despite presenter Mr. Garrett's assertion that the device could match nearly any human vocal feat, Bell Labs envisioned no direct commercialization for the instrument. Instead, the Voder served as an educational exhibit for the 1939 World's Fairs in New York and San Francisco, emerging as a byproduct of underlying research into electrical voice transmission, speech compression, and acoustic analysis. The Voder stood as a public demonstration of speech synthesis rather than an intended consumer product, highlighting the state of early telecommunications research while remaining constrained by its steep operational learning curve.

精炼阅读摘要

1939年贝尔电话实验室展示了世界上首台电子语音合成器Voder，该设备依靠人工对滤波器与能量源进行精密机械控制来模拟人类发声，其本质是贝尔实验室语音压缩与电声传输基础研究的科普示范品，而非直接商业化产品。 声学架构与表达表现力 (0:18) Voder的核心技术构架由两个能量源（嘶嘶声/噪声源与蜂鸣声/音高源）和十组滤波电路组成，二者结合共产生二十个独立的声学分量以组装人类语音。在现场演示中，操作员Helen Harper通过调整控制面板，演示了同一句“She saw me”在不同语境下的重音转移（分别强调主语、宾语或动词）。此外，通过对参数的简单调整，该装置可从标准音高切换至高音及深沉的低音区，还能在持续元音输出中加入颤音调制，将生硬的电蜂鸣转化为具备歌唱质感的音调，完成音阶测试与《友谊地久天长》等唱段表演。 严苛的操作协调与培训门槛 (3:36) Voder并非自动发声系统，其语音输出极度依赖操作人员精细的肢体协调与节奏把控。据操作员Miss Harper介绍，仅合成“concentration”这一个单词，操作者就必须连续组合出13种不同的声音，同时手腕横杆需完成5次上下移动，脚踏板的位置也需根据所需语气调整3到5次。贝尔电话实验室总共选拔了约320名候选人，但历经长达一整年的每日不间断严苛训练后，最终仅有28人获得认证成为专业操作员。Garrett先生指出，合格操作员必须具备一种极其罕见的特殊天赋组合。 拟音能力与非商业化定位 (4:40) 该系统在声学模拟上展现了跨语言与跨物种的适应度。Garrett声称Voder几乎能再现人类声音所能完成的任何表达，演示中不仅清晰发出了法语日常短语“Parlez-vous français?”，还朗读了基础字母表，并模拟了牛鸣与猪哼等动物叫声。针对该设备的商业落地前景，Garrett澄清贝尔电话实验室并不打算将其推向直接的商业市场；Voder研制的首要目标是作为1939年纽约世博会与旧金山金门国际博览会的科普展示项目，其实质是实验室在语音压缩、电声传输与声学分析等通信底层技术研究中所衍生的外显成果。 Voder展示了早期电子语音合成的高度灵活性与声学可行性，但其高度复杂的纯手工操控方式限制了其实际应用。作为1939年世博会的展示样本，它明确服务于公众科学教育与母体通信技术的概念验证，未形成独立的商业化路线。

Source quotes

[4:40] Mr. Garrett: The Voder can do practically anything that can be done with a human voice. [3:36] Mr. Garrett: In operating the Voder, a girl needs a peculiar combination of particular talents which is not too common.

Original transcript

[0:18] there are 10 filter circuits in the [0:20] voter and combined with the two energy [0:22] sources they give a total of 20 separate [0:25] components to be used in building up [0:27] speech sounds but now let's have Mr [0:30] Garrett and Miss Harper actually show us [0:32] what the voter can do with these 20 [0:35] separate sounds well we've heard the [0:37] voter make a word and by combining words [0:40] of course we get a sentence for example [0:42] Helen will you have the voters say she [0:44] saw [0:48] me that sounded awfully flat how about a [0:51] little expression say the sentence in [0:53] answer to these questions who saw you [0:56] she saw me whom did she see she saw [1:00] me well did she see you or hear you she [1:04] saw me now so far you have only heard [1:07] the voters speak in one voice but the [1:09] voter has other voices which he can use [1:12] when Miss harpper makes a simple [1:13] adjustment in his mechanism Helen will [1:16] you have the voter say Greetings [1:18] everybody greetings [1:20] everybody now will you have him repeat [1:22] that in a high voice [1:25] GRE and now in his best space greting [1:31] when a boy's voice changes he's never [1:33] quite sure whether it's going to be a [1:34] tenor or a Bas and the voter being still [1:37] a comparatively young man also has his [1:40] moments of [1:41] uncertainty let's hear him recite Mary [1:43] had a little [1:54] lamb gosh B you sounded awfully dismal [1:57] snap out of it let's hear your laugh [2:03] a minute ago I said the voter was still [2:04] a comparatively young man but that last [2:07] laugh of his makes me wonder whether I [2:09] might not have been wrong do you really [2:11] feel as old as you sound yes [2:16] I well I certainly feel sorry for you [2:20] now perhaps you noticed there was a [2:21] peculiar quality in that voice you've [2:23] not heard before that was the vat we [2:27] noticed a little while ago that a vowel [2:29] sustained without inflection didn't [2:31] sound human it sounded this [2:34] way now if we put fat into that same [2:37] tone we get [2:40] this and that as you noticed is really a [2:43] singing tone yes V it canot only talk [2:46] but he can sing test your voice for us [2:54] voter well if you've quite recovered [2:57] from your mood of depression of a few [2:58] minutes ago suppose you sing a song for [3:01] us will you [3:03] yes well how about o Lang Z [3:32] I'm sure that if I sang that song I [3:34] wouldn't sound as good as the voter Miss [3:36] Harper I've been wondering whether our [3:38] listeners realize how many motions you [3:40] have to make in the production of a [3:41] single word can you give us some idea [3:44] well for example in producing the word [3:47] concentration on the voter I have to [3:49] form 13 different sounds in succession [3:52] make five up and down movements of the [3:54] wrist bar and Vary the position of the [3:56] foot pedals from three to five times [3:58] according to what expression I want the [4:00] voter to give the word and of course all [4:03] this must be done with exactly correct [4:05] timing about how long did it take you to [4:07] become an expert in operating the voter [4:10] it took me about a year of constant [4:11] practice this is about the average time [4:14] required in most cases how many girls [4:16] are there who can operate the voter the [4:19] company tried out about 320 girls and [4:22] out of this number 28 girls finally [4:24] became expert operators well that seems [4:27] like a very small number just why is [4:29] that [4:30] I can answer that Dr Carwell by saying [4:32] that in operating the voter a girl needs [4:34] a peculiar combination of particular [4:37] talents which is not too common Mr [4:40] Garrett does the voter speak any foreign [4:42] language oh yes the voter can talk [4:44] practically any language that it's [4:46] operator can talk well suppose we try a [4:48] little French on the voter can he say [4:50] paru [4:53] F splendidly [4:56] done you see doctor the voter can do [4:59] practically anything that can be done [5:01] with a human voice well that certainly [5:03] covers a lot of ground Mr Garrett I know [5:05] people who can imitate animals can the [5:07] voter do that too oh pretty well for [5:10] instance here's voter's imitation of a [5:16] cow and here's a [5:22] pig have you given the voter any real [5:25] education yet just what do you mean by [5:27] that Dr Carwell well for examp example [5:30] does he know his alphabet I'll let you [5:32] judge that for [5:36] yourself [5:37] F well that certainly proves voters [5:41] versatility and now I'd like to ask just [5:43] what see for the voter Mr Garrett how [5:46] can it be practically applied well Dr [5:49] Carwell we don't anticipate any [5:51] commercial use for the voter it was [5:53] built for an educational exhibit for the [5:55] New York World's Fair and San Francisco [5:57] Exposition however it is a b product of [6:00] developments that we are going on and [6:02] doing in the Bell Telephone [6:04] Laboratories will you please make the [6:05] voters safe for our Eastern listeners [6:08] good evening radio audience good evening [6:11] radio audience