Voicemail Transcription Accuracy via Voice Training Data

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing voice mail transcription methods lack accuracy due to unfamiliarity with the sender's voice patterns, leading to inconvenient and emotion-depicting limitations in both voice and text messages.

Innovation Solution

A communication device with a processor and storage medium that detects audio signals and transmits voice training data to a voicemail system, enabling the transcription module to learn and accurately generate text representations of voice mail messages using voice patterns, inflections, and emphasis.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If voice mail messages are transcribed by a local device, then text message convenience is improved, but transcription accuracy deteriorates due to unfamiliarity with sender's voice patterns

Engineering Contradiction:

Improvetext message convenienceVSAvoidtranscription accuracy

Core Design Contradiction:

Ease of operationVSMeasurement precision

Solution Approach 1:

A remote server acts as an intermediary between the local device and the voicemail system. The server receives audio signals from the voicemail system, applies voice training data from multiple users to improve transcription accuracy, and returns the transcribed text to the local device. This mediator resolves the contradiction by providing accurate transcription without requiring the local device to analyze voice patterns directly.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

Voice training data is collected and processed in advance before actual transcription is needed. The system pre-processes voice patterns from multiple users and stores them as training data on the remote server. When transcription is required, the pre-prepared training data is already available to improve accuracy immediately, rather than requiring real-time voice pattern analysis.

Inventive Principle:

Principle #10Preliminary action

2Loss of information

If voice mail messages are used, then emotional depiction is improved, but receiver convenience deteriorates due to the need to listen to audio

Engineering Contradiction:

Improveemotional depictionVSAvoidreceiver convenience

Core Design Contradiction:

Loss of informationVSEase of operation

Solution Approach 1:

The system replaces the mechanical process of listening to audio with an automated transcription process. The remote server automatically converts audio voice mail messages into text format using voice training data, allowing receivers to read messages instead of listening. This substitution maintains emotional depiction through accurate transcription while significantly improving receiver convenience.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system provides self-service by automatically transcribing voice mail messages without requiring manual intervention from the receiver. The remote server autonomously processes the audio, applies appropriate voice training data, generates the transcription, and delivers it to the local device, freeing the receiver from having to listen to audio messages.

Inventive Principle:

Principle #25Self-service

3Ease of operation

If text messages are used, then receiver convenience is improved, but emotional depiction deteriorates compared to voice mail messages

Engineering Contradiction:

Improvereceiver convenienceVSAvoidemotional depiction

Core Design Contradiction:

Ease of operationVSLoss of information

Solution Approach 1:

The system enhances standard text messaging by substituting simple text generation with an intelligent transcription process. The remote server analyzes audio voice mail messages and converts them into text that preserves emotional nuances through accurate transcription, combining the convenience of text with the emotional richness of voice communication.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

4Measurement precision

If voice training data from multiple users is transmitted, then transcription accuracy is improved, but data transmission volume increases

Engineering Contradiction:

Improvetranscription accuracyVSAvoiddata transmission volume

Core Design Contradiction:

Measurement precisionVSQuantity of substance

Solution Approach 1:

The system extracts only the essential voice training data from multiple users and transmits only this extracted information to the remote server. Rather than transmitting complete audio recordings or excessive data, the system identifies and transmits only the critical voice pattern information needed for accurate transcription, reducing data volume while maintaining accuracy.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The system applies different voice training data from multiple users selectively based on the specific transcription task. Rather than using a uniform approach, the system chooses appropriate voice training data from the available pool, optimizing transcription accuracy for each specific voice mail message while managing data transmission efficiently.

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS9628603B2Voice mail transcription

Publication Date: 2017.04.18 LENOVO SWITZERLAND INTERNATIONAL GMBH

AI summary

For voice mail transcription, a method is disclosed that includes detecting a communication device communicating an audio signal from the communication device to a voicemail system and transmitting data selected from the group consisting of text message data generated from the audio signal and voice training data to the voicemail system.