History of speech synthesis: the era of Electrical solutions

In 1850, German physicist and physiologist Hermann von Helmholtz presented his resonator theory . He noticed that vowel sounds have different resonant frequencies (formants). These formants are formed when a sound wave travels from the vocal cords to the lips. Electrical solutions Some waves are reflected from the speaker's lips and go to the recipient, and some returns to the source. The scientist suggested that the human vocal tract can be represented as a sequence of resonators.

At the beginning of the 20th century, attempts began to implement such a model based on electrical components. The first synthesizer of this type was developed by the physicist John Stewart. His diagram ( publishedin the journal Nature), included an electric buzzer for modeling bundles and a pair of inductive-capacitive resonators. They emulated the physical processes that occur with the sound in the throat. device could produce sounds consisting of two formants. These are a few simple vowels as well asdiphthongs. But that was where his possibilities ended.

The first electric synthesizer capable of reproducing speechappearedlater - in the 1930s. It was developed by Homer Dudley of Bell Laboratories. The company was working on avocoder at the time.- a tool for compressing speech and saving frequency resources of radio links in telephone networks. The idea was to transmit key parameters instead of the subscriber's voice. A special decoder was installed on the receiving side, which, according to these parameters, reconstructed and reproduced the sound. Dudley realized that with a few modifications, the vocoder could be turned into a full synthesizer. This is how the VODER - Voice Operating Demonstrator system appeared .

The device was presented to the general public at the New York World's Fair in 1939. The VODER design includedtwo sound sources: a tube noise generator for “deaf” phonemes, and an oscillator for “voiced” ones. There were also ten parallel-connected bandpass filters - they made up the resonance control unit. The operator operated the system using a hand-held keyboard, wrist strap and foot pedal.

During the demonstrations, the apparatus spoke different languages, sang and answered questions with different intonations. But it took the operator years of training to unleash the system's potential. Shortly after the premiere of VODER, World War II broke out and Bell Labs had to curtail further development of the synthesizer. However, Electrical solutions the knowledge gained during the work on the project was used by Homer Dudley to create technology for encrypting telephone conversations.

Speech synthesizers on spectrograms

In 1946, the acoustic spectrograph was invented . And the idea arose - to use spectrograms to control speech synthesizers. One of the first such devices was presented by L. Schott, an American engineer from Bell Labs. Electrical solutions used a linear light source translucent spectrographic templates with varying degrees of transparency. Special photocells installed opposite the lamp recorded changes in the illumination level and generated control signals for bandpass filters. Homer Dudley used exactly the same filters for his VODER

other development in this area has presented a group of US scientists led by physicist Franklin Cooper ( Franklin to Cooper Cooper ). Their optical system isPattern Playback - Modulated 120Hz pitch harmonics by reading images on a moving transparent tape. The visual information was transmitted to an oscillator that turned it into sound. Electrical solutions system resembled Soviet optical synthesizers - "Nivoton" and "Variofon" - on which they wrote music for cartoons. However, Pattern Playback was originally "sharpened" for the generation of human speech and was able to reproduce entire sentences.