Follow TV Tropes


Synthetic Voice Actor

Go To

manymistakesweremade asked: What voice program do you use to voice The One Whose Shape Was Snatched
Rata: Rude? That’s his voice
— Ask on Rank10YGO's Tumblr blog

Writing a movie? Need a mechanical-sounding voice for your robot? Want to make an episode of your comedy web animation Stylistic Suck? Voice actors are so difficult. What if there was something easier?

A Synthetic Voice Actor (or a synthespian) is a synthetic voice program that voices a character. It's not used a lot, especially when union rules would make that difficult. It's usually used for extremely robotic voices, or a No Celebrities Were Harmed version of Stephen Hawking. It more commonly springs up in Abridged Series and machinima, partly to get extra voices, and partly because of Rule of Funny. When used against human actors, it tends to make the speaker seem inhuman — in more serious works, it's used for threatening robotic characters, usually. Compare the computer voice on the Enterprise (real person) to AUTO (not a real person).


Popular application programs include Speechelo, Talkia and, the last of which is browser-based and free to use on a limited basis. This trope may not apply to programs like MorphVOX Pro, which are merely voice changers.

Compare Machine Monotone, Virtual Celebrity, Auto-Tune. Subtrope of Computer Voice due to how it is painfully obvious to the listener that the line was voiced by a machine.



    open/close all folders 

  • While the original Japanese version of Dragon Ball Super's Auto Magetta is voiced by Naoki Tatsuta, when the series was dubbed in English by Funimation, the character was instead voiced by a text-to-voice program where he spoke entirely in onomatopoeia for robotic sounds.
  • The episode title announcer in Serial Experiments Lain was a Macintosh program named PlainTalk (often falsely called "Whisperer" because of its "Whisper" voice mode).


    Live-Action TV 
  • The Cylons in the 1978 original Battlestar Galactica series spoke this way (human actors run through a synthesizer).
  • Doctor Who: Averted in all instances.
    • The BBC originally considered doing this for the Daleks, but with 1963 technology, they could have done only 45 seconds of dialogue this way, so they used a human voices filtered through a ring modulator. It's actually pretty easy to duplicate the Dalek voices. Record your voice with Dalek speech-patterns, over-amplify it to add clipping distortions (this step is the one people often forget or don't know about, including, occasionally, the actual BBC effects people), then run the results through a ring modulator plugin using 20-40Hz for the frequency of the modulation.
    • The initial creepy sing-song voice of the Cybermen in "The Tenth Planet" was created by (human) voice actors imitating the glitchy speech cadences of the first ever 'singing' computer, IBM 704. The BBC did the Cyberman voices for most of the 1960s by having a human actor use an electrolarynx (an artificial throat-vibrating device for people who had lost their larynx to cancer or injury - they're rarely heard nowadays, but a prominent fictional user is Ned in South Park). The 2009 revival version of the Cybermen simply had an actor's voice run through a ring modulator with a different setting to what was used for the Daleks.
    • The Daleks, also, do not have mechanical voices, only voices that sound mechanical. A truly mechanical voice would probably be one-note-just-like-this, but Daleks have a cadence to their voices, and they also go "EX-TER-MI-NATE! EX-TER-MI-NAAAAATE!" with each intonation rising in pitch and volume. They look like tin cans, but they have some powerful emotion inside them.
    • The Mechonoids in "The Chase" have very peculiar voices created by cutting up a tape of a human voice actor until it is just phonemes, a sort of analog Vocaloid. They are nearly incomprehensible.
  • The person with Locked-In Syndrome in Scrubs.
  • In The Mandalorian season finale, Luke Skywalker's voice is this, using the audio equivalent of deepfake to compose new lines for the character using archival samples of Mark Hamill's voice in the 70s and 80s, as the real Hamill's voice changed significantly since the Return of the Jedi days. This was carried forward with The Book of Boba Fett.

  • "Internet Friends" by Knife Party has the woman who threatens to kill the listener over a Facebook block. In addition to "Internet Friends", there is also the narrator in "Micropenis" and an irritated "fan" in "Superstar" that is reminiscent of their Unpleasable Fanbase.
    Oh my God. What the fuck is this disco shit? What happened to the dubstep?
  • Passenger's "X-Star" from the album For All Mankind uses the SoftVoice Text-to-Speech system.
  • The Radiohead song "Fitter Happier" is "sung" by Mac PlainTalk.
  • Ken Leavitt-Lawrence, better known as MC Hawking, who uses a text-to-speech program to do parody-gangster rap under the guise of Stephen Hawking himself.
  • Kraftwerk may have been the pioneers of using this trope in music.
  • Erasure's cover of "Video Killed the Radio Star" was "sung" by the keyboardist's laptop, since the human singer refused to sing it.
  • Vocaloid:
    • The entire point of the series. Though they still have a human voice source.
    • Played straight with Defoko, the default voice of the similar program UTAU, who is sourced from a program called AquesTalk.
  • Apoptygma Berzerk's "Kathy's Song" has the chorus sung by the Mac text-to-speech Kathy voice. Of course, the song is essentially an exchange between the singer and his computer. It's awesome.
  • My. Name. Is. Skrillex.
  • Camper Van Beethoven's version of "Sisters Of The Moon" by Fleetwood Mac has a text-to-speech program reciting the lyrics (and also throwing in seemingly arbitrary quotes from Pindar, William Shakespeare, and This is Spın̈al Tap). As with Erasure's "Video Killed The Radio Star" cover, this was done because no one in the band wanted to sing it.
  • The voice of the fictional singer Lumi of the Genki Rockets is thought to be either synthesized or a composite of Rachel Rhodes and Nami Miyahara.
  • Elise's singing voice in Sound Horizon's "Märchen" was created using the Hatsune Miku Vocaloid software, with Revo's reasoning being that it made sense for a Creepy Doll to have an artificial voice. Her speaking voice, on the other hand, is provided by Fujita Saki (aka, the original source for Miku's voice).
  • Assemblage 23's "Automaton" uses a vocoder (which he rarely uses) to complement the song's subject.
  •'s "Change", "8 Bits", "I Love 64", "Unknown", "Amnesia" and its thematic Sequel Song "World of Promises" have heavily processed vocals to emulate this effect, though the actual singer is human. Loosely connected songs (i.e. "Remember") and songs from the point of view of the main character, Black, tend to use no or minimal processing.
  • Beck's "Ghettochip Malfunction (Hell Yes)" has a deep synthetic voice echoing some of the lyrics.
  • Overlaps with Celebrity Voice Actor: Stephen Hawking lent his voice to "Keep Talking" on Pink Floyd's "The Division Bell."
    It doesn't have to be like this. All we need to do is make sure we keep talking.
  • This parody of "Cruise" by Florida Georgia Line was done using the scorewriter programs Melody Assistant and Virtual Singer, the latter of which is Exactly What It Says on the Tin.
  • Speaking of bro-country, a synthesized voice (either Microsoft Sam or a soundalike) recites a Title Drop at the beginning of Chase Rice's "Ready Set Roll".
  • The HMKids song "Adeptus Mechanicus" is a Machine-Cult prayer recited by a Text-To-Speech program, accompanied by pipe organs, drums, and electric guitar.
  • The Eurobeat artist April, of "Hanami" and "The Magic I Feel", appears to be a completely synthetic voice.
  • The Plogue virtual instrument plugin chipspeech contains a bank of recordings of various ancient speech synthesisers, which can be played via a keyboard like a musical instrument. Each voice is assigned a VOCALOID-esque character:
    • Dandy 704, a rogueish Steampunk cyborg voiced by the 1961 'singing' computer IBM 704.
    • Otto Mozer, a cyborg Mad Scientist voiced by the 1975 chip the TSI S14001A (the soundchip used to provide the voice of Evil Otto in the game Berzerk as well as voices from Ghostbusters (1984) and Impossible Mission.). The name is a homage to both Evil Otto and the designer of the chip, Professor Forrest S. Mozer[1].
    • Lady Parsec, a Silk Hiding Steel robot queen/despot voiced by the TI-99/4A plug-in speech synthesizer module for the game Parsec.
    • Lady Parsec HD, based on a mixture of Lady Parsec's master audio files and new phonemes provided by a similar-sounding voice actress, which is suspected to be the wife of the head developer, David Viens.
    • Bert Gotrax, a foul-mouthed robot boy voiced by the Votrax SC-01 chip used in QBert.
    • Dee Klatt, a GenderBender cyborg voiced by DECTalk (Steven Hawking's voice unit).
    • Spencer AL2, an Energy Being voiced by the SP0256-AL2 chip used in a number of 80s kit speech synthesisers including the Intellivoice Intellovision attachment and the Magnavox Odyssey voice attachment.
    • Terminal 99, a mysterious computer that may have absorbed its programmer's souls and is worshipped by them as a god, voiced by a TI 99/4A plugin synth module.
    • VOSIM, a robot companion scrapped for being The Unintelligible who tries to make friends with others, voiced by a standard DAC. Notably, his voice is more synth like than the others.
    • Dr. CiderTalk, a cybernetic tech CEO with questionable ethics, voiced by MacinTalk 1.0.

  • Electra the Mornington Crescent Computer, as occasionally featured on I'm Sorry I Haven't a Clue, who sounds like a Satnav, if a Satnav was capable of sarcasm, getting confused, and falling in love with Jeremy Hardy.


  • Texas Instruments' Speak n' Spell toys deserve a special mention due to the fact that they have a very early TI synthesizer, model TMC0280, that the Periphery Demographic loves to abuse.
  • Several early V-Tech educational computer systems also has optional speech modules that can be purchased separately and added to the toy to provide speech.

    Video Games 
  • Animal Crossing uses a simple voice synthesizer to speak each letter of the text, very quickly. Characters even add the proper inflection if a sentence ends with a question mark. It's fairly unintelligible if you try to listen to it, but in the original Japanese it works a lot better, in fact, almost perfectly, since the Japanese language used a series of fixed pronouns compared to western languages which phonemes has to be combined to form words.
  • Several Apple ][ sound cards include a speech chip, if not a DIP socket for a speech chip upgrade. Usually these take Votrax SC-1 or SSI-263 chips.
  • Atari used Texas Instruments speech synthesis chips in several of their mid-1980s Arcade Games, including their Star Wars arcade game and its sequels, the Indiana Jones and the Temple of Doom Licensed Game, Paperboy and Gauntlet.
  • beatmania IIDX 15: DJ TROOPERS used Microsoft Sam for the "Enemy Plane Appoach" voice in the music used for Attract Mode and some of the menus.
  • Stern's Berzerk is likely the Ur-Example for video games. It was either this game, or Midway's Wizard of Wor; both were released the same year (1980).
  • Commodore 64:
    • The 64 and 128 has an add-on module called the Magic Talker that gave it a synthetic voice with a limited vocabulary of 235 words.
    • The Commodore 64 also has the Currah Speech 64 add-on, This device does a much better job compared to the Magic Talker because unlike the Magic Talker, the device actually uses phoneme-splicing instead of pre-programmed words.
  • The merchant in Crypt Of The Necrodancer doesn't "speak", per se, but he "sings" along to the background music by way of a synth voice track layered over the rest of the soundtrack, which can be used to locate his store (and its useful goods). Just be careful if the singing seems off-key.
  • Justified in Dreamfall Chapters. Kidbot and Shitbot both use text-to-speech programs for their voices. Given that they are robots who, In-Universe, use voice synthesizer chips, this makes sense.
  • In the laserdisc arcade game Firefox, this is how "Mitchell Gant" (Clint Eastwood's character from the movie) is portrayed.
  • In Gran Turismo 4, the song "My Precious" is sung by a MacinTalk-style computer voice.
  • The House of the Dead series used these in the first two games. Which, of course, made for some hilariously emotionless bits of drama. As well as the infamous "Suffer like G did?"
  • Impossible Mission on the Commodore 64: "Another visitor? Stay awhile. Stay FOREVER!" note 
  • The Intellivision's Intellivoice module.
  • As an April Fools' Day joke, Four Leaf Studios announced that Katawa Shoujo will be fully voiced thanks to voice synthetisation technology.
  • The Remnant Psyches in Killer7 all use text-to-speech voices. In the original Japanese, they're speaking slightly-hard-to-make-out-but-still-understandable Gratuitous English, but in the English dub, everything they say is run through extra audio filters to make them more incomprehensible.
  • Microsoft Sam and Mary appear chanting the name and motto of the Wii party game, Let's Tap in the game's theme tune.
  • All of the robot bosses in Metal Gear Solid: Peace Walker are voiced by a VOCALOID specially designed for speech rather than singing. When the game was released on the PSP, players even had the option to use the same VOCALOID software to customize voice lines for Metal Gear ZEKE, although the service has since shut down.
  • Moonbase Alpha has this for players. This leads to notorious griefing possibilities.
  • NSFW ~ Not a Simulator For Working uses various synthetic voices to create the "porn sounds".
  • Opening Night and American Girls Premiere, two games that are intended to introduce young children to the concepts of theatre and make their own plays, both use the same text-to-speech system. Suffice to say, most kids used it to create some really really weird plays. (Especially since they lacked a word blacklist.)
  • Peasant's Quest from The Brothers Chaps uses this for Trogdor's voice. Word of God is that they used the Software Automatic Mouth program on an Apple ][ with a supported sound card to generate the source speech.
  • The Portal series is an aversion of this, since its Artificial Intelligences are all voiced by humans (though the voices of GLaDOS and the turrets are heavily edited in post-production).
    • GLaDOS was originally going to be voiced by a text-to-speech program, but Valve wanted her to go through a Villainous Breakdown where she became progressively more emotional, which computer programs couldn't do. The results of running the text of GLaDOS's lines through a text-to-speech program were actually used to coach actress Ellen McLain with regard to giving GLaDOS her distinctive voice in Portal.
    • However, fan-made mods can't afford Ellen's voice (though, once people did think of asking her), hence, they use voice synthesizing programs for their GLaDOSes and AIs (dependent on the story for the mod). However, Ellen does make exceptions and has provided her service in the voice of GLaDOS for a much smaller fee for fan projects in the past.
    • This only becomes apparent in the mod Portal: Prelude, whose main gimmick is that the test supervisors were human, rather than an AI. However, the voices were done in a text-to-speech program because the creators were French and weren't very fluent in spoken English. They couldn't find voice actors in the timeframe in which they wanted to develop the game. This becomes especially awkward towards the end of the game after GLaDOS is turned on for the first time, and she uses her sound files from the main Portal game, mixed with synthesized voices. So we have a robot that sounds more human than the humans do, and uses two personalities at once. For those of you keeping track: Portal's supervisors are robots voiced by humans imitating robots imitating humans, while Prelude's supervisors are humans voiced by robots imitating humans.
  • Q*bert uses a Votrax SC-01 voice synthesizer chipnote  to supply the voices of the various characters. Unfortunately, the chip used couldn't really produce coherent phrases, so the designers decided to go the other direction and have the characters speak a Starfish Language. Word of God is that the programmer in charge of programming the speech was too frustrated with the chip's API due to how hard it was to get it to say what he wanted properly, and simply decided to give up and have Q*Bert speak gibberish.
  • The 2013 Rise of the Triad reboot uses the same synthesizer as Moonbase Alpha for its text chat in multiplayer. Many of the memes follow suit.
  • Safe Cracker has two safes (a voice-activated one, and a vault at the endgame), which operate using voice clips spoken by the PlainTalk voice, "Victoria".
  • Zimos in Saints Row: The Third speaks with an AutoTuned electronic voice box, although Alex Désert does provide his voice.
  • Averted with Williams Electronics' infamous Sinistar, which used a CVSD-basednote  HC-55516 to play back digital audio- Sinistar's speech were digitized recordings of radio personality John Doremus.
  • The main menu theme of Slap City is sung by a Vocaloid.
  • The Tandy Color Computer has the Speech/Sound Program Pak which is apparently using the same synth chip as the Currah Speech 64.
  • Deliberately used badly in Time Fcuk—the main character's voice is barely even comprehensible, and definitely doesn't sound human. It's not quite certain why this is—perhaps the Rule of Scary, or a deliberate attempt to "anonymize" him?
  • Tomodachi Life has this for every character, and lets you customize the voice using various sliders. The difficulty of getting the text-to-speech working in languages other than Japanese was responsible for slowing down the game's western release. The first game was not localized at all due to the DS not being able to properly run an English text-to-speech system.
  • The fan-made Yukkuris in the Touhou franchise tend to be voiced like this, most often by the AquesTalk text-to-speech software.
  • The voice of Byte from Tron 2.0 is actually a voice of MacinTALK.
    • It wasn't the first video game tie-in for TRON to use this. The standing arcade game and Tron Solar Sailer for the Intellivision (using the Intellovoice attachment) also used synthetic voice sound clips in-game. Given the franchise in question, it was a rather appropriate use!
  • FreeSpace 2, the open-source version, has this feature, usually taking installed Windows text-to-speech soundbanks, to provide in-game voices. The feature is optional.
  • HYPII, the AI helper from Hypnospace Outlaw uses the the OS-level text-to-voice software to "speak" to the player. This includes the name the player gives (with an extra box to write it phonetically if HYPII has trouble with it).
  • The ZX Spectrum has the Cheetah Sweet Talker, of which those who own one would attest that it sound anything but sweet- its synthesis was even compared to the voice of a Dalek. The Cheetah Sweet Talker also uses phoneme splicing, and at its heart is a General Instruments SP0256-AL2 speech chip, the same chip used in the Intellivision Intellivoice module and the Magnavox Odyssey voice module.
  • In Project SEKAI, all of the Virtual Singers are voiced by their respective NT software. Datamining shows that the developers use the casts voice actors as a reference for their dialogue.
  • The STAFF Bots of Five Nights at Freddy's: Security Breach use computer generated voices. Since these are the most common type of animatronic found in the Pizzaplex it's meant to give the impression of them being mass produced and not being an animatronic designed for a specific purpose, like the game's four main animatronics.
  • Some music from Shin Megami Tensei III: Nocturne contains vocals by Albert, a text-to-speech voice from Mac computers.
  • After much speculation by the fans regarding the voice of ADA, the computer assistant in Satisfactory, Jace, one of the community managers at Coffee Stain Studios, revealed that the voice was completely generated using Google's WaveNet speech synthesis engine. Specifically, US English Voice C.

    Web Animation 

    Web Original 

    Western Animation 
  • The BOTS Master used a synthesized voice for the evil Corp's robots, while the BOYZZ were all voiced by real actors. This was done to show that the BOYZZ were more human than their soulless corporate adversaries. To be exact, Creative Labs'note  Dr. Sbaitso program is used. To make things more interesting, Sbaitso actually uses Creative Text Assist (itself already used for several movies) for it's speech backend.
  • Steve the Disabled Professor (a Stephen Hawking stand-in) in Family Guy also used Macintalk. Ironically, three episodes also had animated cameo appearances by the real Stephen Hawking.
  • Puter, the Bat-Computer from The LEGO Batman Movie, is voiced by Siri. As noted under Real Life below, Siri's voice isn't entirely synthetic, but the distinction between being voiced by Siri and being voiced by Susan Bennett is there.
  • Compuhorse from Spliced does have a credited voice actor (Patrick McKenna, who also voices Two-Legs Joe), but his speech is filtered through a synthetic program to make him sound more "computer-y".
  • Transformers: Animated: Perceptor's voice is completely synthetic, and probably so to bring to mind Professor Stephen Hawking. One of the writers has suggested that he "deleted his emotions and personality" to make room for more data, though (like many an "emotionless" character), he certainly seems to have both, if understated.
  • H.E.L.P.eR. from The Venture Bros. is voiced by something called "Soul-Bot", which processes Chris McCulloch's "eep!" noises into electronic beeps and boops.
  • AUTO from WALL•E. As if there weren't enough Apple references in the movie already, the voice is Mac OSX's own MacinTALK. Similarly, M-O's "Foreign Contaminant" is provided by PlainTalk.
  • War Planets (AKA: Shadow Raiders), a CGI series created by Mainframe Entertainment (of Reboot famenote ) had Princess Tekla's companion device voiced by MacinTalk.
  • The title robot in Whatever Happened to... Robot Jones? originally used MacinTalk Junior, credited as "Himself", but later switched to a modulated human's voice, and they even redubbed the earlier episodes with the real actor.

    Real Life 
  • NOAA Weather Radio went to all-synthetic voices in the late '90s, with most stations having one male and one female synthetic voice, and a Spanish-speaking voice is used where needed. This is also done for Canada's Weatheradio service, with stations alternating between English and French voices. Select TV stations will use synthetic voices for their Emergency Alert System tests as well, and Canada exclusively does for their equivalent Alert Ready warnings.
  • Surprisingly, Siri averts this - voice actress Susan Bennett (and her international counterparts) spent five hours a day for four weeks voicing seemingly random words for ScanSoft (the company behind Siri's technology), who then sliced them up to form the phonemes and intonations that the search engine uses.
  • The Software Automatic Mouth, or SAM, by Don't Ask Software, provides a CPU-driven speech synthesizer to any machine that has a PSG chip. The Apple II version supported several PSG-based sound cards made for the Apple II (and was a pack-in software for several cards), and the C64 version drives the SID chip directly. The Atari 8-bit version used the POKEY chip (another PSG chip on Atari 8-bit computers that can also be found in several arcade machines), but had other limitations due to the design of the system (the ANTIC chip had to be disabled, as do interrupts- this means that the screen has to be blanked while generating speech). The PC and base Apple II version can even try to generate speech from the internal beeper by treating it as a rudimentary PSG. MacInTalk was a rebranded version of this software that Steve Jobs personally commissioned Don't Ask Software to produce for the Mac (which went on to voice AUTO in WALL•E, among many other roles). Microsoft Voice/Speech (of which Microsoft Sam is a part of) is also undoubtedly its spiritual successor if not direct spinoff or even direct continuation of the line.
  • Stephen Hawking, famously, whose distinctive Machine Monotone voice is probably the Trope Codifier inspiring many of these examples. Many people who are unable to talk for medical reasons have followed in Hawking's lead, which is much easier and more affordable now due to technological advancement. Averted with famed heavy metal guitarist Jason Becker (who suffers from ALS and can only move his eyes and some facial muscles), as he and his father devised an eye movement system that allows him to convey messages faster than he could with a computer.
  • In the 1980s, Texas Instruments released the Vocaid, a modified Speak & Spell that allowed people who were physically unable to talk to speak with a robot voice. Pre-loaded voice samples and button overlays allowed the user to call 911, request assistance from caregivers, spell letters & numbers, and even speak when playing card games. The Vocaid is very much an outdated product now, but it has attracted some attention in modding communities in recent years.
  • Pulitzer Prize winning film critic Roger Ebert used a computer to communicate after he lost his voice due to thyroid cancer. Software engineers began using clips from DVD commentaries that he recorded to synthesize his own voice, which he demonstrated on The Oprah Winfrey Show, but his variant wasn't fully completed before his death.
  • Used by the blind in screen reader software, which translate a computer screen's contents into synthesized speech (or braille). Favorites include DEC talk, of Moonbase Alpha fame, originally a hardware synthesizer from the 80s, and Eloquence, used in JAWS among other platforms.
  • Very common in Japanese Let's Play culture on Nico Nico Douga circa the late 2000s and early 2010s, as players wanted to comment on the game in real-time but were uncomfortable having their real voices recorded, and would thus use text-to-speech programs to read out their commentary. This started off with AquesTalk of Yukkuri fame, but later software such as Voiceroid and CeVIO Creative Studio emerged with more realistic-sounding voices; some of said programs came associated with characters of their own, and it became popular for the resulting LPs to be done "in-character" in a fashion similar to that of Vocaloid culture (to which said software has deep links to). The universal prevalence started to die down somewhat in the late 2010s due to the rise of more Western-esque YouTube "talk show" culture in Japan, particularly with the Virtual YouTuber phenomenon offering alternate means of privacy and interaction, but even on YouTube many privacy-oriented uploaders still elect to have talk-to-speech software do the commentary for them.
  • The comedian Lee Ridley aka Lost Voice Guy, who suffers from cerebal palsy. One line in his act is that the only impression he can do is Stephen Hawking, but he's really good at it.