# US 2024/0203402 A1 · Voice processing using photonic quantum computing

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

- Number: US 2024/0203402 A1
- Document ID: US20240203402A1
- Status: Published application
- Kind: A1
- Application number: 18/068,873
- Filed: 2022-12-20
- Published: 2024-06-20
- Earliest claimed: 2022-12-20
- Assignees: Bank of America Corporation
- Applicants: Bank of America Corporation
- Inventors: Prashant Anna Bidkar; Ankit Upadhyaya; Prashant Khare; Mani Arora
- Web page: https://patentgrep.com/patent/US20240203402A1

## CPC

- G10L 15/16: using artificial neural networks
- G06N 10/60: Quantum algorithms, e.g. based on quantum optimisation, quantum Fourier or Hadamard transforms
- G06V 10/88: Image or video recognition using optical means, e.g. reference filters, holographic masks, frequency domain filters or spatial domain filters
- G10L 15/02: Feature extraction for speech recognition; Selection of recognition unit
- G10L 15/1815: {Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning}
- G10L 15/1822: {Parsing for meaning understanding}
- G10L 15/22: Procedures used during a speech recognition process, e.g. man-machine dialogue
- G10L 25/06: the extracted parameters being correlation coefficients
- G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

A system and method for voice processing using photonic quantum computing. A method includes receiving voice data represented by classical binary bits. The voice data is converted to a first photon beam including converted voice data represented by photonic quantum bits. The first photon beam is split into split photon beams. The split photon beams are received by a quantum neural network, which includes quantum neural network clusters. The converted voice data is processed by processing the split photon beams. Each of the split photon beams is processed by a respective neural network cluster in parallel. Processing the converted voice data includes extracting voice features. Word embeddings are created based on the voice features. Sentences are determined based on the word embeddings.

## Claims

### Claim 1

An apparatus comprising:
a converter system configured to:
receive voice data, wherein the voice data is represented by classical binary bits; and
convert the voice data to a first photon beam comprising converted voice data,
wherein the converted voice data is represented by photonic quantum bits;
a photonic quantum computing system communicatively coupled the converter system, the photonic quantum computing system comprising:
a beam splitter configured to:
receive the first photon beam; and
split the first photon beam into a plurality of split photon beams; and
a photonic quantum processor coupled to the beam splitter and implementing a quantum neural network, wherein the quantum neural network comprises a plurality of quantum neural network clusters, and wherein the photonic quantum processor is configured to:
receive the plurality of split photon beams; and
process the converted voice data by processing the plurality of split photon beams, wherein each of the plurality of split photon beams is processed by a respective neural network cluster in parallel, and wherein processing the converted voice data comprises:
extracting a plurality of voice features from the converted voice data;
creating word embeddings based on the plurality of voice features;
determining a plurality of sentences based on the word embeddings;
determining first correlations between each word and all other words of each sentence based on the word embeddings;
identifying a context of each sentence based on the first correlations;
determining second correlations between each word of each sentence and all words of a previous sentence based on the word embeddings;
determining an intent of the plurality of sentences based on the first correlations and the second correlations; and
sending the plurality of sentences and the intent as a second photon beam.

### Claim 2 (depends on 1)

The apparatus of claim 1, wherein the voice features comprise a frequency, a pitch, an amplitude, a zero crossing rate, a root mean square energy, a mel spectrogram, a band energy ratio, and a spectral centroid.

### Claim 3 (depends on 1)

The apparatus of claim 1, wherein creating the word embeddings comprises determining matrix representations of respective words.

### Claim 4 (depends on 1)

The apparatus of claim 1, wherein determining the first correlations between each word and all other words of each sentence based on the word embeddings comprise calculating matrix product of a matrix representation of each word and matrix representations of all other words of each sentence.

### Claim 5 (depends on 1)

The apparatus of claim 1, wherein determining the second correlations between each word of each sentence and all words of a previous sentence based on the word embeddings comprise calculating a matrix product of a matrix representation of each word of each sentence and matrix representations of all words of the previous sentence.

### Claim 6 (depends on 1)

The apparatus of claim 1, wherein the converter system is further configured to:
receive the second photon beam;
convert the plurality of sentences to a plurality of converted sentences, wherein the plurality of converted sentences are represented by classical binary bits; and
convert the intent to a converted intent, wherein the converted intent is represented by classical binary bits.

### Claim 7 (depends on 6)

The apparatus of claim 6, wherein the photonic quantum processor is further configured to:
send the plurality of converted sentences and the converted intent to an application system.

### Claim 8

A method comprising:
receiving voice data, wherein the voice data is represented by classical binary bits;
converting the voice data to a first photon beam comprising converted voice data, wherein the converted voice data is represented by photonic quantum bits;
splitting the first photon beam into a plurality of split photon beams;
receiving the plurality of split photon beams by a quantum neural network, wherein the quantum neural network comprises a plurality of quantum neural network clusters; and
processing the converted voice data by processing the plurality of split photon beams, wherein each of the plurality of split photon beams is processed by a respective neural network cluster in parallel, and wherein processing the converted voice data comprises:
extracting a plurality of voice features from the converted voice data;
creating word embeddings based on the plurality of voice features;
determining a plurality of sentences based on the word embeddings;
determining first correlations between each word and all other words of each sentence based on the word embeddings;
identifying a context of each sentence based on the first correlations;
determining second correlations between each word of each sentence and all words of a previous sentence based on the word embeddings;
an intent of the plurality of sentences based on the first correlations and the second correlations; and
sending the plurality of sentences and the intent as a second photon beam.

### Claim 9 (depends on 8)

The method of claim 8, wherein the voice features comprise a frequency, a pitch, an amplitude, a zero crossing rate, a root mean square energy, a mel spectrogram, a band energy ratio, and a spectral centroid.

### Claim 10 (depends on 8)

The method of claim 8, wherein creating the word embeddings comprises determining matrix representations of respective words.

### Claim 11 (depends on 8)

The method of claim 8, wherein determining the first correlations between each word and all other words of each sentence based on the word embeddings comprise calculating matrix product of a matrix representation of each word and matrix representations of all other words of each sentence.

### Claim 12 (depends on 8)

The method of claim 8, wherein determining the second correlations between each word of each sentence and all words of a previous sentence based on the word embeddings comprise calculating a matrix product of a matrix representation of each word of each sentence and matrix representations of all words of the previous sentence.

### Claim 13 (depends on 8)

The method of claim 8, further comprising:
receiving the second photon beam;
converting the plurality of sentences to a plurality of converted sentences, wherein the plurality of converted sentences are represented by classical binary bits; and
converting the intent to a converted intent, wherein the converted intent is represented by classical binary bits.

### Claim 14 (depends on 13)

The method of claim 13, further comprising:
sending the plurality of converted sentences and the converted intent to an application system.

### Claim 15

A non-transitory computer-readable medium storing instructions that, when executed by one or more processors, cause the one or more processors to:
receive voice data, wherein the voice data is represented by classical binary bits;
convert the voice data to a first photon beam comprising converted voice data, wherein the converted voice data is represented by photonic quantum bits;
split the first photon beam into a plurality of split photon beams;
receive the plurality of split photon beams by a quantum neural network, wherein the quantum neural network comprises a plurality of quantum neural network clusters; and
process the converted voice data by processing the plurality of split photon beams, wherein each of the plurality of split photon beams is processed by a respective neural network cluster in parallel, and wherein processing the converted voice data comprises:
extracting a plurality of voice features from the converted voice data;
creating word embeddings based on the plurality of voice features;
determining a plurality of sentences based on the word embeddings;
determining first correlations between each word and all other words of each sentence based on the word embeddings;
identifying a context of each sentence based on the first correlations;
determining second correlations between each word of each sentence and all words of a previous sentence based on the word embeddings;
an intent of the plurality of sentences based on the first correlations and the second correlations; and
sending the plurality of sentences and the intent as a second photon beam.

### Claim 16 (depends on 15)

The non-transitory computer-readable medium of claim 15, wherein creating the word embeddings comprises determining matrix representations of respective words.

### Claim 17 (depends on 15)

The non-transitory computer-readable medium of claim 15, wherein determining the first correlations between each word and all other words of each sentence based on the word embeddings comprise calculating matrix product of a matrix representation of each word and matrix representations of all other words of each sentence.

### Claim 18 (depends on 15)

The non-transitory computer-readable medium of claim 15, wherein determining the second correlations between each word of each sentence and all words of a previous sentence based on the word embeddings comprise calculating a matrix product of a matrix representation of each word of each sentence and matrix representations of all words of the previous sentence.

### Claim 19 (depends on 15)

The non-transitory computer-readable medium of claim 15, wherein the instructions, when executed by the one or more processors, further cause the one or more processors to:
receive the second photon beam;
convert the plurality of sentences to a plurality of converted sentences, wherein the plurality of converted sentences are represented by classical binary bits; and
convert the intent to a converted intent, wherein the converted intent is represented by classical binary bits.

### Claim 20 (depends on 15)

The non-transitory computer-readable medium of claim 15, wherein at least one of the one or more processors is a photonic quantum processor.
