Skip to content
vocalfry.org
EN
AI voices

Why does AI voice have vocal fry?

The most likely reason is that AI voices learn from real people, and real people creak at the ends of sentences. A neural text-to-speech model that learns “how a sentence ends” picks up the falling pitch and the creak together. Researchers treat creak as something to control in synthetic speech, demonstrating that synthesized creak can be modelled and adjusted.

Updated · 11 sources

Illustration of a smart speaker and laptop showing a synthetic voice waveform ending in creaky pulses

What is actually happening

Modern text-to-speech systems are trained on hours of recorded speech, increasingly including spontaneous, conversational speech because it sounds more natural than read text. Spontaneous speech contains a lot of creak, especially at phrase ends. A 2023 paper at the International Congress of Phonetic Sciences, from a team at KTH in Stockholm, built neural speech synthesis with controllable creaky voice for exactly this reason: creak is part of natural speech, and it is hard to handle.

The same group’s 2024 Interspeech paper, CreakVC, goes further. It represents creak frame by frame, using a creak detector (CreaPy) combined with a voice-conversion model (FreeVC), so the amount of creak in converted speech can be dialled up or down. The paper names voice-interactive systems and media production as possible uses.

Waveform of a vocal fry recording, showing separate spikes instead of a smooth band
Fry looks like a row of separate spikes in a waveform. A speech model trained on recordings like this learns that pattern too. Waveform drawn by vocalfry.org from a CC BY 4.0 research recording.

A plausible mechanism for sentence-final creak

Put those findings together and a likely chain appears:

  1. The training recordings contain sentence-final creak.
  2. The model learns prosody, the rise and fall of a sentence, and creak travels with the falling pitch at the end.
  3. Default controls in most products adjust speed, pitch or style, but rarely have a dedicated “creak” setting.
  4. So creak appears as a by-product, more in some voices than others.

This is an inference from research, not a vendor statement. We found no AI company explaining why its own voices creak, and we have not run listening tests on specific commercial products, so we make no claim about any particular assistant.

Why people notice it

Creak lowers perceived pitch and adds irregular pulses, and listeners notice it mainly through that low pitch, according to a 2023 Applied Sciences study and the 2026 McGill work. The 2014 PLOS ONE study shows listeners attach strong judgments to fry in human voices. People bring those reactions to synthetic voices too, which is part of why a creaky AI voice gets commented on at all.

Can you remove it?

In research, yes: CreakVC shows creak can be controlled. In products, it depends. Sonarworks, which sells audio tools, notes that AI voice plugins learn fry and grit from large sets of recordings, and that the result depends on the plugin, the quality of the input and the target voice; they do not perfectly reproduce real fry. In practice, your options are usually to pick a different voice, change the speaking style if the tool allows it, or edit the audio afterwards. Be wary of anything promising “one-click fry removal”.

Not all AI voices creak

Whether you hear fry depends on the model, the recordings it learned from and the specific voice. Some voice libraries even sell creak as a feature: ElevenLabs’ voice library has a “croaky” category with voices described as having intentional vocal fry (we saw this as a search listing). Singing synthesis is similar; VOCALOID6 users add fry through parameter editing, as covered on vocal fry in singing.

One common mix-up: Sam Altman, CEO of OpenAI, is often discussed for his own vocal fry. That is his natural voice and has nothing to do with how ChatGPT’s synthetic voices sound. See Sam Altman’s vocal fry.

Further reading: CreakVC (Interspeech 2024, PDF) · Controllable creaky voice synthesis (ICPhS 2023, PDF)

Questions people ask

Why do AI voices sound creaky?

A likely reason: neural text-to-speech learns from recordings of real people, and real speech is full of sentence-final creak. The model learns that creak belongs at the end of sentences along with falling pitch. Researchers at KTH have built systems specifically to control how much creak synthetic speech has, which shows it is a known side effect.

Can you remove vocal fry from an AI voice?

Sometimes. Research systems such as CreakVC can turn creak up or down, and some voice tools let you choose a different voice or adjust style. Whether a particular product can remove it depends on that product; there is no universal switch.

Is there a vocal fry voice generator?

Some AI voice libraries offer intentionally creaky voices; ElevenLabs lists a “croaky” category. We explain the phenomenon here and do not offer a generator.

Do all AI voices have vocal fry?

No. It depends on the model, the training recordings and the chosen voice. Some voices creak noticeably, others hardly at all.

Sources

Review dates are shown for each reference. Labels distinguish full text, abstracts, metadata and search snippets; each label describes the evidence available for that reference.

  1. Lameris et al., CreakVC, Interspeech 2024 (PDF) full text
  2. ICPhS 2023 — Neural speech synthesis with controllable creaky voice (PDF) full text
  3. Sonarworks — Can AI voice plugins generate realistic vocal fry and grit? (2025) full text
  4. ElevenLabs Voice Library — Croaky voices search snippet
  5. Anderson, Klofstad, Mayew & Venkatachalam (2014), PLOS ONE 9(5): e97506 full text
  6. Devaraj et al. (2023), Applied Sciences 13(7): 4186 full text
  7. Brown & Sonderegger (2025), Journal of Phonetics — Canadian English-French bilingual creaky voice metadata
  8. Jeanne Brown — Author summary of the 2025 bilingual production study full text
  9. Brown & Clayards (2026), Laboratory Phonology — Acoustic and social biases in creaky voice perception full text
  10. New Scientist — Vocal fry is more common in men, actually (14 May 2026) full text
  11. YouTube — Zyla Zo., How to add vocal fry with Vocaloid6 AI (tutorial) metadata

Advertisement