# US 12,585,895 B2 · Communication channel quality improvement system using machine conversions

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 12,585,895 B2
- Document ID: US12585895B2
- Status: Granted
- Kind: B2
- Application number: 18/814,983
- Filed: 2024-08-26
- Published: 2025-10-30
- Granted: 2026-03-24
- Earliest claimed: 2024-04-25
- Assignees: Height Ventures, LLC
- Applicants: Height Ventures, LLC
- Inventors: Ken R. Davis; Fraser M. Smith
- Web page: https://patentgrep.com/patent/US12585895B2

## CPC

- G06F 40/58: Use of machine translation, e.g. for multi-lingual retrieval, for server-side translation for client devices or for real-time translation
- G06F 40/103: Formatting, i.e. changing of presentation of documents (automatic justification G06F40/189; automatic line break hyphenation G06F40/191)
- G06F 40/166: Editing, e.g. inserting or deleting
- G06F 40/263: Language identification
- G06F 40/44: Statistical methods, e.g. probability models
- G06F 40/51: Translation evaluation
- G10L 15/1822: {Parsing for meaning understanding}
- G10L 21/013: Adapting to target pitch
- G10L 13/00: Speech synthesis; Text to speech systems
- G10L 15/26: Speech to text systems (G10L15/08 takes precedence)
- G10L 2021/0135: {Voice conversion or morphing}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING

## Abstract

Technology is described for adding inserted message data into machine learning modified communications. The method can include receiving message data from a sender to be sent to a recipient. Another operation may be estimating a conversion time to convert the message data to a second language that is different than a language of the sender, using a machine learning translation service. A message insert may be provided or identified that approximates the conversion time. The message insert to be reproduced for the recipient may be sent. The message data may be sent to the recipient as converted to the second language.

## Claims

### Claim 1

A method for assessing quality of a translation pipeline that uses machine learning, comprising:
receiving message data from a sender to be sent to a recipient;
correcting, in real-time, errors in the message data using a normalization service to provide a corrected message;
converting the message data to a second language, using a machine learning translation service, wherein the message data is corrected at least one of grammatically, mechanically or semantically;
sending the message data to the recipient to be reproduced for the recipient;
measuring a time length of task completion requested in the message data; and
determining whether the time length of task completion represents a successful or unsuccessful translation, wherein the time length of task completion is measured using at least one of a task completion service, software instrumentation of an application the recipient is working on, hardware instrumentation of an application the recipient is working on, or APIs (application program interfaces) integrated in an application the recipient is working on.

### Claim 2 (depends on 1)

The method as in claim 1, further comprising identifying faster task completion times that represent successful translations.

### Claim 3 (depends on 1)

The method as in claim 1, further comprising identifying slower task completion times that represent unsuccessful translations.

### Claim 4 (depends on 1)

The method as in claim 1, further comprising:
measuring a number of clarification requests from the recipient associated with the message data;
setting a quality threshold for the number of clarification requests; and
determining a translation to the second language is lower quality when the quality threshold for the number of clarification requests is reached.

### Claim 5 (depends on 1)

The method as in claim 1, further comprising:
presenting the recipient with a quality feedback survey upon completion of a task; and
determining the quality of a translation based on responses to the quality feedback survey.

### Claim 6 (depends on 1)

The method as in claim 1, further comprising determining quality feedback based in part on whether a customer is able to perform requested actions.

### Claim 7 (depends on 6)

The method as in claim 6, wherein the requested actions comprise at least one of granular actions, events, groups of events, or complex actions.

### Claim 8 (depends on 1)

The method as in claim 1, further comprising determining quality feedback based in part on whether a customer is able to take requested actions in a web application or client application.

### Claim 9 (depends on 1)

The method as in claim 1, further comprising:
tracking a session length of a session for communication between the sender and recipient; and
determining that the communications are improving when the session length is decreased as compared to similar sessions.

### Claim 10 (depends on 9)

The method as in claim 9, wherein the session is categorized as at least one of a subject, a topic, or a type.

### Claim 11 (depends on 9)

The method as in claim 9, further comprising comparing a session comprising the converted message data to a similar session comprising raw message data to identify one or more system deficiencies.

### Claim 12 (depends on 1)

The method as in claim 1, wherein faster task completion times may be identified that represent successful translations or slower task completion times may be identified that represent unsuccessful translations.
