LipConfirm
Whispers too faint to transcribe reliably can still help reduce typing errors. This approach offers a new middle ground between conventional typing and full dictation.
U.S. Patent No. 12,737,540 · 30 claims · Issued September 15, 2026
Mobile typing can demand attention before anything goes wrong. You may already anticipate the next mistake by autocorrect—and the interruption of stopping to fix it. Typing more precisely or constantly checking the text takes attention too, can feel frustrating, and can pull you away from what you want to say.
Dictation offers another route, but speaking aloud takes energy and effort. You may be tired, want to avoid disturbing others, or prefer privacy and not having your message overheard. Even when you are alone, you may simply not feel like speaking an entire message. LipConfirm explores a new middle ground: keep typing while accompanying the words with barely audible articulation to help reduce annoying autocorrect errors. That faint articulation can feel much less effortful than dictating aloud.
If you have ever faintly articulated words while reading to yourself or working through instructions, the action may already feel familiar. LipConfirm’s founding idea grew from that observation and a personal frustration with autocorrect. You already know the word you intend; your typing and faint articulation can express it together, in your own rhythm, without learning a new communication technique.
A faint whisper may be too quiet for reliable dictation, but its acoustic signal does not have to carry the message by itself. Combined with your typing, even incomplete clues can help distinguish the intended word as you compose. The goal is more forgiving typing: less pressure to hit every key perfectly, fewer interruptions to repair text, and more attention left for what you want to say.
LipConfirm allows faint whispering or, alternatively, completely silent lip movements as the accompanying signal. Faint-whisper assistance uses the microphone; silent lip assistance uses the camera to follow mouth movements while you type.
See how typing and a faint whisper can reinforce one another. These preset examples illustrate how supporting evidence can help identify an intended word—the mechanism behind the proposed ongoing typing mode. You can also explore silent lip assistance.
restaurant
Generate candidates from typing first, then use lip or whisper evidence to rescore and select.
Typing only · no auxiliary evidence
Type the word to see its candidates.
Waiting for the mistyped word.
Illustrative simulation. No microphone or camera is accessed.
Typing stays primary in this simulation. Whisper and lip cues guide correction rather than provide a standalone dictation or lipreading result. A single articulation can supply both cues in a combined implementation; combined capture is not simulated here.
Even LLM-informed autocorrect can struggle with garbled input or a typo that produces another valid word, such as a slip from “love” to “live” or “this” to “thus”. A quiet cue together with your typing adds evidence of what you intended, beyond what spelling and context alone can establish.
A faint whisper combined with your typing can be more useful than the same ambiguous whisper alone. Partial clues, such as syllable rhythm or a distinguishing vowel, can help infer the intended word when evaluated together with your typing. LipConfirm uses that evidence without requiring a reliable standalone transcript. Whisper assistance needs no camera.
The same supporting role can be supplied by silent articulation. LipConfirm can evaluate visible lip shapes and movements corresponding to words (visemes) alongside your typing. Ongoing lip assistance requires camera capture during the assisted typing period; an individual requested correction can use a brief capture. Faint-whisper assistance offers a camera-free alternative.
Faintly articulate words as you type, across a phrase or a whole message, so supporting evidence is available without first noticing an error. You can also use assistance selectively or add a cue afterward to reconsider a word. Continuous support is the primary use case; requested correction remains another option.
A very faint whisper may fail to register or may produce the wrong words at normal typing distance. Speaking louder or moving the phone toward your mouth can help. But raising your voice can quickly approach regular speaking volume, defeating the purpose of staying quiet. Moving the phone closer can also interrupt typing.
LipConfirm combines that faint whisper with the letters you type, so the audio does not have to identify the intended word on its own. A whisper too faint or ambiguous for reliable dictation can still provide useful clues alongside your typing. The aim is to let you keep typing with comfortable, barely audible articulation, without needing to make the whisper clear enough for standalone dictation.
The point at which dictation becomes unreliable varies by device and setting.
LipConfirm combines that uncertain signal with your typing to help resolve the intended word as you compose. The whisper need not be clear enough to identify the word on its own. For a silent alternative, camera capture can supply lip evidence throughout an assisted typing session or for an individual requested correction.
Potential accessibility applications. LipConfirm may also make correction easier for people who can type but find repeated backspacing or precise cursor placement difficult. For users who can articulate words but have difficulty producing audible speech, silent lip movements could provide another way to clarify an intended word. The choice of cue can be matched to the user’s abilities and preferences.
Typing errors can leave a word garbled or turn it into another valid word. A faint whisper or silent lip movement accompanying your typing can add evidence to help identify what you intended. See what the cue contributes in selected examples below.
U I O sit side by side on a QWERTY keyboard. That makes pairs such as love/live, this/thus, if/of, and pick/puck easy to mistype. These tiny slips can be frustrating because a real word may pass a spelling check, and context does not always rescue the mistake. Faster or less precise typing can also leave a word garbled, making the intended spelling harder to recover from the typed letters alone.
| # | You meant | Typed by mistake |
|---|---|---|
| 01 | love | live |
| 02 | this | thus |
| 03 | if | of |
| 04 | shirt | short |
| 05 | pick | puck |
| 06 | restaurant | restrant |
| 07 | avocado | abcado |
| 08 | appointment | apointmrnt |
| 09 | prescription | prescriotn |
| 10 | accommodation | acxomodatn |
| # | You meant | Typed by mistake |
|---|---|---|
| 11 | I love my life here. | I live my life here. |
| 12 | I left it in the car. | I left it on the car. |
| 13 | I like the shirt dress. | I like the short dress. |
| 14 | We discussed the renovation. | We discussed the renovstn. |
| 15 | Please bring the coat. | Please bring the boat. |
| 16 | The team is ready. | The tea is ready. |
| 17 | Please send the confirmation. | Please send the confrmatn. |
| 18 | I’ll go with the latte. | I’ll go with the latter. |
| 19 | It’s a casual relationship. | It’s a causal relationship. |
| 20 | What about Wednesday? | What about wrfnesdy? |
In the valid-word examples, both versions are grammatical but mean different things. A language model can favor one, yet still miss your intent.
A cue need not reveal the whole word to help resolve it. Even faint whispers often carry clues to syllable count, rhythm, critical vowel or consonant sounds (phonemes), and timing. Optional camera capture can add visemes, visible lip shapes associated with speech sounds, such as closure or rounding, and the movements between them. LipConfirm combines the available cues with the typing to help infer the intended word; audio and lip evidence can also reinforce one another.
In “I love my life here,” a faint “uh” vowel cue can favor love over live, pronounced “liv.” The same vowel contrast helps separate thus from this. Within each pair, the consonants and syllable count match. The typing narrows the choice; the vowel adds evidence context may lack.
If you type abcado but whisper avocado, the surviving letters still provide part of the word’s structure. The whisper can add vowel sounds and syllable rhythm that the garbled spelling does not preserve. LipConfirm can evaluate both together, including in a multimodal model that infers the word directly from the combined inputs. The whisper need not identify the word on its own.
If you type “The tea is ready” but articulate team, the lips close for the final “m.” A captured closure timed to that word ending can support team over tea. LipConfirm can weigh that visible cue alongside the shared typed letters; a resting mouth closure alone would not establish the word.
A captured initial “k” sound can favor coat over boat. In the other direction, boat begins with a “b” made by closing and releasing both lips, while “k” is formed farther back in the mouth. A captured lip closure and release at the word’s start can support boat when evaluated together with the typing.
When casual is articulated with three syllables, its rhythm differs from the usual two syllables of causal. The vowel and middle consonant sounds also differ. LipConfirm can weigh those clues against the nearly identical typed letters. Pronunciation varies, so syllable count is supporting evidence rather than a fixed rule.
These words differ by the neighboring “i” and “o” keys. Their vowel visemes can also differ: the “ih” in if can have a different visible shape from the more open vowel in a fully articulated of. LipConfirm can weigh that shape and its transition into the final consonant against the typing, with whisper audio when available. The visible contrast depends on pronunciation and articulation.
These walkthroughs illustrate how cues can support correction. Pronunciation and capture quality determine which evidence is available. When the combined evidence is insufficient, LipConfirm can retain the typed word or offer alternatives.
LipConfirm is designed to bring typed input and supporting whisper or lip evidence into existing keyboard, autocorrect, spell-checking, and language-model infrastructure. Technical demonstrations and IP discussions available. Desktop prototypes cover both whisper-assisted and lip-assisted correction.
In a user-enabled session, the microphone captures faint articulation alongside keystrokes. LipConfirm associates the acoustic evidence with the text being entered and evaluates them together as input arrives, allowing for timing differences between articulation and typing. A reliable standalone transcript is not required. Silent lip evidence can serve the same supporting role through camera capture.
Rerank candidates: Start with suggestions generated from typing, then use whisper or lip evidence to rescore or extend those candidates.
Infer with a multimodal model: Evaluate typing and whisper or lip evidence together in a learned model to infer the intended word. The word need not already appear in a separate autocorrect list.
The combined evidence can update a suggestion, confirm a word, or correct a valid word that was typed unintentionally. Confidence controls matter for familiar words as well as garbled input: when the evidence is insufficient, the system can retain the text or offer alternatives.
A faint whisper often retains useful clues to syllable count and rhythm, critical vowel or consonant sounds (phonemes), and sound timing. LipConfirm evaluates the available acoustic evidence together with the original typing, including garbled text. The whisper need not identify the word independently: a partial cue can help resolve the intended word when combined with the typing. A desktop demo of whisper-assisted correction is available.
The capture path must preserve that weak signal. Depending on the system, speech-detection gates or audio filtering can exclude faint utterances before they reach recognition. A user-enabled typing session or requested capture can provide a path for evaluating the retained acoustic evidence with the text. A blank dictation result alone does not identify which stage failed.
Phonetic background: research on whispered speech documents vowel resonances and sound durations despite the absence of normal vocal-fold vibration. This helps explain why incomplete acoustic evidence can remain informative; it does not establish a minimum usable volume for LipConfirm.
Whisper remains the primary auxiliary cue. If you prefer not to whisper at all, LipConfirm can use the camera to track your silent lip movements throughout an assisted typing session or during a brief requested capture. It is designed to recognize visemes, the transitions between them, and visual syllable patterns such as the rhythm of mouth opening and closing. These provide alternative visual evidence to evaluate alongside the original typing.
The table below shows examples of visemes detected by the current LipConfirm desktop demo.
| Viseme category | Visible pattern and example sounds |
|---|---|
| CLOSED Bilabial | Both lips meet, as for /p/ in pat, /b/ in bat, and /m/ in mat. |
| LABIO | The lower lip contacts the upper teeth, as for /f/ in fan and /v/ in van. |
| SPREAD | The lips spread for an “ee” sound, /iː/, as in see. |
| OPEN | The mouth opens for an “ah” sound, /ɑ/, as in father. |
| ROUNDED | The lips round or protrude for “oo,” /uː/, as in food, or “oh,” /oʊ/, as in go. |
| MID | A relaxed, moderately open mouth shape for “ih,” /ɪ/, as in sit or if. |
LipConfirm can combine these shapes, their transitions, and visual syllable patterns with the typed letters. For example, a final CLOSED shape can support team over tea; a SPREAD or MID vowel shape can help separate words with “ee” or “ih” sounds.
These references provide background for the visual speech cues and articulation described above. LipConfirm’s preliminary correction results are described separately in the accuracy FAQ.
U.S. Patent No. 12,737,540 issued September 15, 2026, with all 30 claims. The patent addresses methods for combining typing with whisper audio, lip movements, or both to help identify the words a user intends.
Two U.S. continuation applications are pending, seeking additional protection for lip- and whisper-assisted autocorrect. One has been granted Track One prioritized examination.
PCT/US2026/035963 is pending. The USPTO, acting as the International Searching Authority, issued a written opinion with positive novelty, inventive-step, and industrial-applicability findings for all 30 claims. The international search report contains no category X or Y citations.
You may prefer typing because speaking aloud feels effortful, could disturb others, or could make your message easier to overhear. The faint articulation that feels comfortable alongside typing may be too quiet for reliable dictation at normal phone distance. LipConfirm adds the evidence in your actual typed input, so the same uncertain whisper can help resolve a word without having to identify it independently. This can be useful even as dictation improves.
LipConfirm is showing that combining faint whisper cues with typed input can help identify the intended word, including when the typing is garbled. These are early findings from selected desktop demonstrations. A representative accuracy percentage across users, devices, noise levels, and normal phone typing distances has not yet been established. Broader evaluation needs to count both recovered words and unwanted corrections.
No. Familiar words can be mistyped into other valid words, including love/live and this/thus. “I love my life here” and “I live my life here” are both grammatical, so even a strong language model may not settle the user’s intent from context alone. A whisper or lip cue can supply additional evidence; confidence checks determine whether to change the word.
The microphone can capture a faint whisper even when dictation cannot reliably transcribe it. Useful acoustic clues can remain in that audio even if dictation returns no text or the wrong word. LipConfirm uses the captured signal together with your original typing to help infer the intended word, without requiring a reliable standalone transcript. Try the progressively fainter whisper check at your usual typing distance.
Faint articulation can be less audible to people nearby than ordinary speech, but it is not a guarantee against being overheard or recorded. If you articulate a whole message, the audio may contain information about that whole message. Silent lip assistance is an alternative when you do not want to produce audible speech.
For data privacy, the architecture supports local processing and discarding raw audio or mouth-region frames. The host platform determines the implemented data handling.
Here, whisper-level audio means quiet acoustic evidence, including faint, low-amplitude or breathy speech. The aim is to use comfortable, barely audible articulation in your natural typing rhythm while keeping the phone where you normally type. LipConfirm evaluates that evidence with the typed input; a complete, correct standalone transcript is not required. Usable volume depends on capture settings, microphone distance, background noise, and the model.
No. Whisper-assisted correction combines microphone input with typing and does not need a camera. Optional lip assistance can provide a silent cue or reinforce a quiet utterance.
For ongoing lip-assisted typing, the camera needs to capture mouth movements during the assisted period. For an individual requested correction, a brief capture can be enough. Whisper-assisted typing does not require a camera.
Yes. The architecture supports ongoing whisper-assisted typing during an opted-in session. You can faintly articulate each word in your natural typing rhythm, use the extra input only for part of a message, or request help with an individual word. There is no requirement to articulate every word.
The architecture supports associating a new cue with the word just typed and reconsidering the correction using the retained typing evidence. Optional lip assistance can use a brief camera capture requested for that word.
Yes. One quiet articulation can supply audio and lip cues. The architecture combines both with typed evidence and adjusts their influence according to reliability. An unavailable or unreliable channel can contribute less or be omitted.
Desktop demonstrations of whisper-assisted and lip-assisted correction are available. The website demo is an illustrative simulation with preset examples and scores; it does not capture live audio or video. The proposed continuous typing experience builds on these mechanisms. Its effects on accuracy, speed, effort, and user preference still need broader evaluation.
LipConfirm is designed to support on-device correction without sending raw mouth-region frames or audio to the cloud. Raw sensor data can be discarded after feature extraction or scoring. Deployment choices depend on the host platform and model.
The design supports session-based or requested microphone capture, with camera capture only for lip assistance. Ongoing lip assistance uses the camera during the assisted typing period; individual requested corrections can use brief captures. Actual battery use and latency depend on capture duration, sensor settings, recognition models, and the device.
LipConfirm can rescore or extend suggestions generated from typing, or infer a correction through a learned multimodal model that evaluates the inputs together. A proxy-string pathway can turn speech-related clues into spelling-like inputs for an existing correction engine. The interactive demo illustrates late and early fusion as two distinct strategies.
Text-assisted correction can use partial cues, such as syllable patterns and mouth shapes, rather than independently recognizing every word from lips alone. Useful visual evidence still depends on articulation and capture quality. Optional personalization can adapt scoring to a user.
The architecture supports keyboard-based correction on phones, tablets, desktop systems, wearables, and extended-reality devices. Each platform can use the available, user-permitted combination of typed, acoustic, and visual evidence.
A high-confidence result can be inserted or confirmed automatically. When the evidence is uncertain, the system can offer alternatives, retain the current text, or wait for another cue. The user does not need to see the internal candidate scores or camera feed.
Yes. U.S. Patent No. 12,737,540 issued September 15, 2026, with all 30 claims. Two U.S. continuation applications and a PCT international application are pending. One continuation has been granted Track One prioritized examination. See Patent & IP for details.
For IP acquisition, licensing, or a technical demonstration, get in touch.
Email: hello@lipconfirm.com
LipConfirm was conceived and developed by Andre Persidsky, a multidisciplinary engineer and inventor with experience building, patenting, and licensing technology. Andre has previously licensed technology to commercial partners including Spotify, and has led hardware/software health-technology projects in collaboration with external research institutions.