howclose.to
Biology · Updated July 2026Momentum · accelerating ↗

How close are we to giving speech back through a brain implant?Can an implant turn thoughts back into speech?

or, simply: Can an implant turn thoughts back into speech?or, precisely: How close are we to giving speech back through a brain implant?

Implanted electrodes can already turn neural activity into words on a screen; the open questions are speed, vocabulary, and whether it lasts for years.Electrodes read the brain's signals for speech. The race is to make them fast, accurate, and durable enough for daily life.

We are here

Wireless speech trial implants first — Connect-One implanted its first participant with a fully internal 421-electrode BCI designed to test text and synthesized speech. Next up — Two-person safety endpoint scheduled (expected May 2027).

01 · Where we stand

Four tests between here and the goal

Each threshold is a falsifiable claim with a named next test. We move the meter only when a result is public.

Open-vocabulary brain-to-textSay more than fixed phrases✓ Achieved · Aug 2023
100%
Proven byOne participant reached 62 words/minute with 23.8% error on a 125,000-word vocabulary.
Real-time expressive voiceRestore a natural voiceIn progress
60%
Next testAchieve under-10% listener WER on unseen spontaneous sentences with independently controlled pitch and emphasis.
Independent multi-year home useWork every day at homeIn progress
70%
Next testRepeat two-year, 1,000-hour-plus use in several participants using a fully implanted wireless device.
Reproducible clinical benefitWork for many peopleEarly
20%
Next testA prospective multicenter cohort must publish user-level accuracy, complications, explants and daily-use outcomes.
THRESHOLDS — Thresholds for Speech BCI.
Scale
Error-adjusted decoded communication rate: linear scaleError-adjusted decoded communication rate over time, with measured values, projected values, and a goal at 120 correct words/minute.020406080100120Error-adjusted decoded communication rate · correct words/minuteYear20212023.62023.61202420252026GOAL 120 · Sustained independent open conversation at the low end of natural speech15.2 WPM × 74.4% correct; 50-word lab task: 11.3 correct words/minute (2021)15.2 WPM × 74.4% correct; 50-word lab task62 WPM × 76.2% correct; 125,000-word prompted task: 47.2 correct words/minute (2023.6)62 WPM × 76.2% correct; 125,000-word prompted task78 WPM × 75% correct; 1,024-word prompted task: 58.5 correct words/minute (2023.61)78 WPM × 75% correct; 1,024-word prompted task32 WPM × 97.5% correct; self-paced conversations: 31.2 correct words/minute (2024)32 WPM × 97.5% correct; self-paced conversations47.5 WPM × 56.4% correct; direct synthesized-voice evaluation: 26.8 correct words/minute (2025)47.5 WPM × 56.4% correct; direct synthesized-voice evaluation49.7 WPM × 96.5% correct; silent-speech benchmark after day 600: 48 correct words/minute (2026)49.7 WPM × 96.5% correct; silent-speech benchmark after day 600~72 correct words/minute to goal
NOTE — Values multiply reported words per minute by one minus word-error rate; this derived proxy prevents speed records from hiding errors. Tasks are not directly comparable: vocabularies, implants, language models, prompted versus spontaneous speech and text versus voice differ. The candidate file's recommended endpoint is error-free independent home communication; only the 2026 participant demonstrated prolonged independent home use, and its 47.96 point is still a structured benchmark rather than audited free-conversation throughput.
02 · How we got here

The record behind the verdict

Major events set large; context events set small but never hidden. Everything below the TODAY rule is a schedule, not a result.

1973–19971 event1 shown

The Direct-Link Hypothesis

The Direct-Link Hypothesis begins with direct brain-computer link proposed. The result established the next question for the field.

1973
Direct brain-computer link proposedTheory
Jacques Vidal formalized direct brain-computer communication using measurable electrical brain signals and online computing.
Jacques Vidal and UCLA
1998–20083 events2 shown

Single Channels Move Cursors

Single Channels Move Cursors moved the field from locked-in patient controls switch to braingate proves multidimensional control. The results narrowed the next question without closing it.

1998
Locked-in patient controls switchExperiment
A neurotrophic cortical electrode recorded action potentials for months and enabled one woman with ALS to make binary selections.
Philip Kennedy and Roy Bakay
2004
Neural typing reaches three characters
A locked-in participant moved a cursor over a virtual keyboard at only 3 characters per minute using an implanted electrode.
Neural Signals researchers
2006
BrainGate proves multidimensional controlExperiment
A 96-channel array let a person with tetraplegia open email, operate a television and move a robotic hand using imagined movement.
BrainGate consortium
2009–20182 events0 shown

Motor Intent Becomes Language

Motor Intent Becomes Language moved the field from implant drives vowel synthesizer to cursor typing accelerates. The results narrowed the next question without closing it.

2009
Implant drives vowel synthesizer
Two neural channels predicted vowel formant frequencies in real time; the system produced vowel-like sounds, not words.
Frank Guenther and colleagues
2017
Cursor typing accelerates
Three participants reached 6.5, 6.3 and 2.7 correct characters per second with a point-and-click intracortical BCI.
Chethan Pandarinath and colleagues
2019–20223 events2 shown

Neural Decoders Learn Speech

Neural Decoders Learn Speech moved the field from cortical activity synthesizes sentences to handwriting bypass reaches phone speed. The results narrowed the next question without closing it.

2021
Anarthric speech decoded onlineExperiment
One participant decoded a 50-word vocabulary at 15.2 words/minute and 25.6% median word error with language modeling.
David Moses and UCSF
2021
Handwriting bypass reaches phone speed
Attempted handwriting produced 90 characters/minute at 94.1% raw online accuracy in one participant.
Francis Willett and BrainGate
2023–203611 events4 shown

Records Confront Daily Life

Records Confront Daily Life moved the field from percutaneous risks stay visible to six-year follow-up scheduled. The results narrowed the next question without closing it.

2023
Percutaneous risks stay visible
BrainGate safety data included one localized pedestal-site infection treated with oral antibiotics; devices remained investigational.
BrainGate consortium
2023
Large vocabulary reaches sixty-twoExperiment
One ALS participant decoded 125,000 words at 62 words/minute with 23.8% word error in prompted laboratory speech.
Francis Willett and Stanford BrainGate
2024
Seven-year system eventually declinesSetback
A fully implanted switch BCI remained functional over 7 years, but performance declined after roughly year 6.
UMC Utrecht BCI team
2025
Voice streams every eighty milliseconds
A 1,024-word ECoG system synthesized personalized speech in 80-ms increments at 47.5 words/minute in one stroke survivor.
UCSF and UC Berkeley
2025
Instant voice remains error-prone
Closed-loop neural voice synthesis was near-instant, but a neural-only evaluation still had 43.60% word error.
Maitreyee Wairagkar and colleagues
2026
Home use crosses 3,800 hoursDeployment
One man independently used a wired intracortical BCI for 3,801 hours over 19 months, producing 1.96 million words at 56 words/minute.
UC Davis and BrainGate
2026
Wireless speech trial implants firstWe are here
Connect-One implanted its first participant with a fully internal 421-electrode BCI designed to test text and synthesized speech.
Paradromics and University of Michigan
2027
Two-person safety endpoint scheduledExperimentTarget
Connect-One lists May 2027 primary completion for an estimated two participants; the primary outcome is device-related adverse events.
Paradromics Connect-One trial
2032
Six-year follow-up scheduledExperimentTarget
The registry estimates study completion in January 2032 after up to six years of long-term safety observation.
Paradromics Connect-One trial
2036–20361 event0 shown

Events outside the declared eras

Events outside the declared eras begins with ten-year wireless durability target. The result established the next question for the field.

2036
Ten-year wireless durability targetDeploymentTarget
Paradromics designs Connexus for more than 10 years of stable recording; this company goal is unproved, not a scheduled approval.
Paradromics
03 · The data behind the verdict

Why the meters read the way they do

The learning curves and comparisons that justify each threshold's percentage. Every series is measured, with the source event linked in the timeline above.

Closing on conversation

Three curves, one threshold

48 / 120error-adjusted wpm → goal
Speech brain interfaces — closing on conversationRaw speed rose sharply, but the points mix laboratory prompts, voice synthesis and real-world personal use; they must be paired with error and vocabulary. Text-decoding error has fallen in some participants; the 2025 value evaluates direct voice intelligibility and is not comparable to text WER. Large-vocabulary text is demonstrated, but voice systems still often use 1,024-word constrained evaluations and language models shape output. The error-adjusted metric goal is 120 correct words per minute.FasterReported decoding rate · words/minute0.067134120 goal15 words/minute (2021)62 words/minute (2023.6)78 words/minute (2023.61)32 words/minute (2024)48 words/minute (2025)56 words/minute (2026)2021202420252026More accurateReported word error rate · %0.0244926 % (2021)24 % (2023.6)25 % (2023.61)2.5 % (2024)44 % (2025)3.5 % (2026)2021202420252026Richer vocabularyDecoder vocabulary size · words10¹10⁴10⁶50 words (2021)125k words (2023.6)1k words (2023.61)125k words (2024)1k words (2025)125k words (2026)2021202420252026Shared study year axis
speed riseserrors can fallvocabulary expands

NOTE — Together, the panels show why conversation is not one leaderboard: useful systems must get faster, more accurate, and broad enough to express ordinary language.

Use moved from hundreds of supervised conversation hours to thousands of independently initiated home hours in one person.
Use moved from hundreds of supervised conversation hours to thousands of independently initiated home hours in one person.1,000Cumulative conversational or home-use exposure · hoursYear20242026Cumulative conversational or home-use exposure: 248 hours (2024)Cumulative conversational or home-use exposure: 3,801 hours (2026)3,801 hours
NOTE — Use moved from hundreds of supervised conversation hours to thousands of independently initiated home hours in one person.
The current strongest deployment evidence is one participant using the system on 444 of 653 observed days, not a multi-user clinical cohort.
The current strongest deployment evidence is one participant using the system on 444 of 653 observed days, not a multi-user clinical cohort.0100200300400500Independent home-use days · days usedYear2026Independent home-use days: 444 days used (2026)444 days used
NOTE — The current strongest deployment evidence is one participant using the system on 444 of 653 observed days, not a multi-user clinical cohort.
  • 2021 handwriting BCI18 words/minute equivalent
  • 2023 speech decoder62 words/minute
COMPARISON — The 2023 large-vocabulary speech system more than tripled the prior 18-word-per-minute handwriting BCI rate.
  • Prompted vocalized benchmark99.2 % word accuracy
  • Personal-use sentences immediately perfect53.3 % sentences
COMPARISON — Benchmark accuracy was much higher than the strict real-world rate of sentences immediately decoded perfectly.
  • 2024 conversation exposure248 hours
  • 2026 independent home use3801 hours
COMPARISON — Independent use expanded more than fifteenfold between the 2024 and 2026 reports.
  • Typical setup20 minutes
  • Maximum continuous use19 hours
COMPARISON — The 2026 home system required a substantial daily setup despite independent operation afterward.
04 · What it unlocks

If the remaining tests pass

Downstream capabilities, drawn dashed because they depend on results not yet in.

Speech BCIRestored conversation after paralysisGive people with ALS, stroke or spinal injury a faster, more personal way to speak with family and clinicians.Independent digital workTurn intended speech into messages, documents and computer control without relying on residual eye or limb movement.Richer neural prosthesesCombine speech, cursor and eventually sensory feedback in durable assistive systems controlled by the user.
05 · Sources

Where every number comes from

  1. Vidal's direct brain-computer communication proposalpubmed.ncbi.nlm.nih.gov
  2. Kennedy and Bakay locked-in cortical linkpubmed.ncbi.nlm.nih.gov
  3. 2009 review of speech communication BCIspmc.ncbi.nlm.nih.gov
  4. 2021 anarthria speech neuroprosthesisnejm.org
  5. 2023 large-vocabulary intracortical speech BCInature.com
  6. 2023 ECoG speech and avatar neuroprosthesisnature.com
  7. 2024 rapidly calibrating speech neuroprosthesisnejm.org
  8. 2025 streaming brain-to-voice neuroprosthesisnature.com
  9. 2025 instantaneous voice synthesisnature.com
  10. 2026 long-term independent home usenature.com
  11. BrainGate interim safety profilepmc.ncbi.nlm.nih.gov
  12. Paradromics Connect-One first implantuofmhealth.org
  13. Connect-One registered study and completion datesclinicaltrials.gov