---
title: >-
  Google Research Project Relate: Personalized speech recognition for transcription,
  restatement, and text entry
slug: google-project-relate-accessibility
stable_id: 33e321501f83de63
company: Google Research Project Relate
function_code: customer_service
pattern_codes:
  - creator_to_judge
evidence_strength: verified
publication_tier: showcase
freshness: watch
reviewed_at: '2026-08-23'
updated_at: '2026-08-22'
source_quality_summary: 1 peer reviewed, 2 primary; publication outcomes are verified.
caveat_summary: >-
  The study is a small feasibility evaluation, not proof of universal accuracy or
  communication benefit; Google currently accepts no new users.
collections: []
bundle_version: 1.0.0
bundle_fingerprint: sha256:c23c6cc2b88153d008ea8fda928f632ce0011fc2d4c5036672a16e5d895bab93
canonical_url: https://brianletort.ai/transformations/google-project-relate-accessibility
---

# Google Research Project Relate: Personalized speech recognition for transcription, restatement, and text entry

People with non-standard speech need not rely only on listener familiarity, repetition, or text entry.

Function: Customer service. Patterns: Creator to judge. Evidence: verified.

Freshness: watch. Reviewed: 2026-08-23. Updated: 2026-08-22.
Watch status: verify the cited source and deployment condition before reusing this case.

Source quality: 1 peer reviewed, 2 primary; publication outcomes are verified.

## Before

1. **Person with non-standard speech** — Speak, repeat, gesture, or switch to typing when a listener or device does not understand. (control: Human accommodation)
2. **Listener** — Interpret, ask for clarification, or fail to complete the interaction. (control: Listener understanding)

## After

1. **User** — Records at least 500 phrases to create a personalized recognition model. (control: Android, connectivity, and voice stability)
2. **Project Relate** — Transcribes speech, restates it in a synthesized voice, or sends text through a keyboard. (control: Personalized model)
3. **User and listener** — Use the output with active listening, gestures, and clarification. (control: Humans retain meaning and communication control)

## Decision rights

The model proposes a transcript or synthesized restatement; the user decides whether it expresses the intended message.

## Exception path

Users repeat, type, gesture, or ask a listener for clarification when recognition is low; severe or rapidly changing speech can make the model unsuitable.

## Outcomes

- **Feasibility of personalized ASR use in Ghana** (verified): General ASR can have 78-89% word-error rates for severely dysarthric speech, with 0-1.2% correctly transcribed sentences in cited prior work → Twenty people with communication difficulties completed a Ghana user study; the study identified feasibility but also access, adaptation, support, and accuracy limitations. The study is a small feasibility evaluation, not proof of universal accuracy or communication benefit; Google currently accepts no new users.

## Executive lesson

Assistive AI is not useful merely because it recognizes more speech; onboarding burden, device access, changing voices, and user control determine whether it increases autonomy.

## Anti-pattern

Treating a transcript as the user's intended message or presenting a closed beta as broadly available.

## Questions for leaders

- Can the user easily reject or correct the output?
- What happens as the user's voice changes?

## Sources

- [Enhancing Communication Equity: Evaluation of an Automated Speech Recognition Application in Ghana](https://doi.org/10.1145/3613904.3641903) — ACM CHI 2024
- [Project Relate: An App for Non-Standard Speech](https://sites.research.google/relate/) — Google Research
- [Project Relate Help Center](https://sites.research.google/relate/help/) — Google Research