Robot voice text to speech: 10 free settings that work
Most “robot voice generators” are a normal voice with one effect on top. Once you know which effect, you can make any robot sound you want for free. We made ten classic robot voices from the same sentence, using FreeTTS voices and one free ffmpeg command each. Press play, pick the one you like and copy its exact settings.
The short answer
- How it works
- A robot voice is a normal text to speech voice plus an audio effect. The voice reads the words; the effect makes it sound like a machine.
- Best free route
- Make the voice in FreeTTS, download the MP3, then run one ffmpeg command from the table below. Audacity works too if you prefer buttons.
- Best starting voice
- Guy or Christopher (English, US) for classic robots, Jenny for a small, high robot. Steady voices give cleaner results.
- Quickest classic robot
- Guy at 0.75x speed, Low pitch, then the robotize command. That is our “classic computer” sample.
- The sci fi villain voice
- Ring modulation with a low tone. Our sample multiplies the voice by a 30 Hz sine wave.
- What FreeTTS does not do
- It has no built in robot effect. The effect step happens after you download.
Robot voice settings at a glance
The FreeTTS voice, speed and pitch for each style, and the effect added afterwards. Commands are in the next section.
| Style | Sounds like | FreeTTS voice | Speed | Pitch | Effect |
|---|---|---|---|---|---|
| Classic computer | The flat, buzzing computer voice from old films | Guy (English, US) | 0.75x | Low | Robotize, then cut the highs |
| Ship AI | A calm onboard computer that is a little too friendly | Aria (English, US) | 1x | Normal | Chorus plus a short echo |
| Ring modulator | The metallic sci fi villain voice | Ryan (English, UK) | 1x | Low | Multiply the voice by a 30 Hz sine wave |
| Walkie talkie droid | A robot speaking over a cheap radio | Christopher (English, US) | 1x | Normal | Robotize, narrow band, 8 bit crush |
| Glitch bot | A broken robot that stutters and crackles | Eric (English, US) | 1.25x | Normal | 4 bit crush plus fast tremolo |
| Deep machine | A huge, slow industrial robot | Christopher (English, US) | 0.75x | Low | Drop the pitch 20%, then robotize |
| Tiny helper bot | A small, cheerful robot sidekick | Jenny (English, US) | 1.25x | High | Raise the pitch 30%, then robotize |
| Space station announcer | A public address voice echoing down a metal corridor | Davis (English, US) | 1x | Normal | Flanger plus a long echo |
| Vocoder choir | Several robots speaking in unison | Guy (English, US) | 1x | Normal | Robotize plus a three voice chorus |
| Old radio robot | A vintage robot from a 1950s radio serial | Sonia (English, UK) | 1x | Normal | Radio band, 6 bit crush, slap echo |
Speed and pitch are the FreeTTS controls: speed runs from 0.5x to 2x, and pitch has Low, Normal and High. Every sample reads the same line: “Greetings, human. I am online, fully charged, and ready to read your text out loud.”
Hear all ten robot voices
Each card has the sample, the settings and the exact command. Run it on your FreeTTS download, saved as voice.mp3.
Classic computer
The flat, buzzing computer voice from old films. Guy (English, US), 0.75x speed, Low pitch. Robotize, then cut the highs.
ffmpeg -i voice.mp3 -af "afftfilt=real='hypot(re,im)*sin(0)':imag='hypot(re,im)*cos(0)':win_size=512:overlap=0.75,lowpass=f=3400" robot.mp3The robotize step throws away the natural rise and fall of the voice and leaves one steady buzz. Slowing Guy down first makes every word land like a machine reading a printout.
Ship AI
A calm onboard computer that is a little too friendly. Aria (English, US), 1x speed, Normal pitch. Chorus plus a short echo.
ffmpeg -i voice.mp3 -af "chorus=0.6:0.9:50|60:0.4|0.32:0.25|0.4:2|2.3,aecho=0.8:0.6:40:0.25" robot.mp3No robotize here, so the words stay clear. The chorus doubles the voice with tiny delays, which gives it that smooth, synthetic sheen.
Ring modulator
The metallic sci fi villain voice. Ryan (English, UK), 1x speed, Low pitch. Multiply the voice by a 30 Hz sine wave.
ffmpeg -i voice.mp3 -filter_complex "[0:a]aformat=sample_rates=24000:channel_layouts=mono[v];sine=frequency=30:sample_rate=24000,volume=8[c];[v][c]amultiply,volume=0.9" -shortest robot.mp3This is ring modulation, the technique the BBC Radiophonic Workshop used for the Daleks. Try frequency=25 for a rougher growl or frequency=50 for a thinner buzz.
Walkie talkie droid
A robot speaking over a cheap radio. Christopher (English, US), 1x speed, Normal pitch. Robotize, narrow band, 8 bit crush.
ffmpeg -i voice.mp3 -af "afftfilt=real='hypot(re,im)*sin(0)':imag='hypot(re,im)*cos(0)':win_size=512:overlap=0.75,highpass=f=500,lowpass=f=2500,acrusher=bits=8:mode=log:aa=1" robot.mp3Cutting everything below 500 Hz and above 2,500 Hz is what makes it sound like it came through a small speaker.
Glitch bot
A broken robot that stutters and crackles. Eric (English, US), 1.25x speed, Normal pitch. 4 bit crush plus fast tremolo.
ffmpeg -i voice.mp3 -af "acrusher=bits=4:mode=log:aa=1,tremolo=f=12:d=0.6" robot.mp3Only 4 bits of detail are left, so the voice fizzes. The tremolo pulses the volume 12 times a second to add the stutter.
Deep machine
A huge, slow industrial robot. Christopher (English, US), 0.75x speed, Low pitch. Drop the pitch 20%, then robotize.
ffmpeg -i voice.mp3 -af "asetrate=24000*0.8,aresample=24000,atempo=1.25,afftfilt=real='hypot(re,im)*sin(0)':imag='hypot(re,im)*cos(0)':win_size=512:overlap=0.75" robot.mp3asetrate lowers the pitch far below what the Low setting can do, and atempo=1.25 puts the timing back so it does not drag.
Tiny helper bot
A small, cheerful robot sidekick. Jenny (English, US), 1.25x speed, High pitch. Raise the pitch 30%, then robotize.
ffmpeg -i voice.mp3 -af "asetrate=24000*1.3,aresample=24000,atempo=0.77,afftfilt=real='hypot(re,im)*sin(0)':imag='hypot(re,im)*cos(0)':win_size=256:overlap=0.75" robot.mp3A smaller robotize window (256 instead of 512) gives a higher, lighter buzz that suits a small robot.
Space station announcer
A public address voice echoing down a metal corridor. Davis (English, US), 1x speed, Normal pitch. Flanger plus a long echo.
ffmpeg -i voice.mp3 -af "flanger=delay=3:depth=4:speed=0.4,aecho=0.8:0.7:120:0.3" robot.mp3The flanger sweeps slowly through the voice and the 120 ms echo puts it in a big hard room. Good for announcements in games.
Vocoder choir
Several robots speaking in unison. Guy (English, US), 1x speed, Normal pitch. Robotize plus a three voice chorus.
ffmpeg -i voice.mp3 -af "afftfilt=real='hypot(re,im)*sin(0)':imag='hypot(re,im)*cos(0)':win_size=512:overlap=0.75,chorus=0.5:0.9:50|60|40:0.4|0.32|0.3:0.25|0.4|0.3:2|2.3|1.3" robot.mp3The chorus stacks three delayed copies on top of the robotized voice. It is the closest free trick to the sound of a real vocoder.
Old radio robot
A vintage robot from a 1950s radio serial. Sonia (English, UK), 1x speed, Normal pitch. Radio band, 6 bit crush, slap echo.
ffmpeg -i voice.mp3 -af "highpass=f=300,lowpass=f=3000,acrusher=bits=6:mode=log:aa=1,aecho=0.8:0.5:20:0.3" robot.mp3No robotize at all. The narrow band and crunchy bits alone make a clean voice sound like it is coming out of an old valve radio.
How to make your own robot voice in five minutes
1. Make the voice. Open FreeTTS, paste your text and pick a voice from the table. Set the speed and pitch shown for your style. Generate the audio and download the MP3. Rename it voice.mp3.
2. Get ffmpeg.ffmpeg is a free, open source audio and video tool that runs on Windows, Mac and Linux. Download it from ffmpeg.org, or install it with your system’s package manager.
3. Run the command. Open a terminal in the folder with voice.mp3, paste the command for your style and press Enter. It writes a new file called robot.mp3. Your original voice file is not changed, so you can try as many styles as you like.
4. Even out the volume. Bit crushing and robotize can change how loud the result is. If it is too quiet, run ffmpeg -i robot.mp3 -af loudnorm final.mp3. This is the same filter we used on the samples above.
5. Tweak one number at a time. Every command has one number that matters most: the 30 in the ring modulator, the bits in acrusher, the 0.8 or 1.3 in asetrate, the 120 in the space station echo. Change that one, listen, repeat.
Robot voices without the command line
Slow, low and steady
Guy or Christopher at 0.5x or 0.75x with Low pitch already sounds flat and mechanical. Good enough for a quick joke or a game prototype.
Audacity
Import the FreeTTS MP3 and stack Change Pitch, Echo and Phaser from the Effect menu. You get close to the ship AI and space station styles.
ffmpeg
The only route where you can copy our settings exactly and get the same sound we did, on any computer.
What makes a voice sound robotic
Four tricks cover almost every robot voice in film, games and music.
Flattening the pitch
Human speech rises and falls all the time. Take that movement away and the brain hears a machine. The robotize command does this: it splits the voice into many small frequency slices, keeps how loud each one is and throws away the rest, leaving a steady buzz that still carries the words. The command is taken directly from the examples in the official ffmpeg documentation for its afftfilt filter.
Ring modulation
A ring modulator multiplies the voice by a steady tone. The result sounds metallic and slightly out of tune. Its most famous use is the Daleks: Brian Hodgson of the BBC Radiophonic Workshop used a ring modulator to create their voice in Doctor Who from 1963. We used 30 Hz; try 25 or 50 in the command to hear how the tone changes.
Vocoders
A vocoder analyses speech into frequency bands and uses them to shape another sound, often a synthesiser. Homer Dudley invented it at Bell Labs in 1937, as a way to synthesise speech, and a related device, the Voder, was shown to the public at the 1939 New York World’s Fair. Our vocoder choir sample fakes it with robotize plus a chorus.
Cheap speakers and missing detail
Cutting the low and high frequencies makes a voice sound like it is coming out of a small speaker or radio, and reducing the bit depth adds crunch. The walkie talkie, glitch bot and old radio styles all use it.
What to watch out for
- Clarity drops with every effect. Robotize, bit crushing and narrow bands all remove information. If listeners must catch every word, use one effect, slow the voice to 0.75x and keep sentences short.
- Expressive voices fight the effect. Voices with lots of breath and emotion keep some of it after processing, which can sound like a person in a tin can rather than a machine.
- Commercial use follows your voice plan. The effect commands are free for anything. The rights to the voice depend on the plan you used to make it; see the commercial use guide.
The fastest way to a good robot voice
A robot voice is two steps: a steady text to speech voice and one audio effect. Start with Guy at 0.75x and Low pitch in FreeTTS, download it and run the robotize command. That gives the classic computer voice in under five minutes, for free.
For the metallic sci fi villain, use ring modulation at 30 Hz. For a friendly ship computer, skip robotize and use chorus with a short echo. Every command on this page is the one we ran, so what you hear is what you get.
Start with the voice
Pick Guy, Christopher or Jenny in FreeTTS, set the speed and pitch from the table and download your MP3. Free to start.
Robot voice questions
A robot voice from text is made in two steps: a steady text to speech voice, then an audio effect. The classic computer sound comes from flattening the voice’s pitch, and the metallic sci fi villain sound from ring modulation with a low tone, around 30 Hz.
Where the facts come from
Our own test
Ten samples generated with FreeTTS voices in September 2026 and processed with the commands shown. The audio files are on this page.
ffmpeg
FFmpeg filters documentation, including the robotize example for afftfilt.
Ring modulation
Ring modulation, on the BBC Radiophonic Workshop and the Daleks.
Vocoder
Vocoder, on Homer Dudley, Bell Labs and the 1939 World’s Fair.
Audacity
Audacity manual, index of effects: Change Pitch, Echo and Phaser.