Speech Data Services

Your Partner in Creating High-Quality Speech Datasets.

Custom Speech Collections

We specialise in creating bespoke speech datasets tailored to your specific requirements. Whether you need data from particular dialects, demographics, or domains, we collect and transcribe speech to enhance your Automatic Speech Recognition (ASR) systems and related applications. We ensure that the datasets align perfectly with your requirements, leading to improved accuracy and performance in your speech-driven technologies.

↓ Contact us to create your dataset.

African Language Datasets Available

Explore our collection of high-quality African-language speech datasets, ready for immediate use in your projects. These off-the-shelf datasets cover a select range of African languages and industries, with new additions coming. Designed to support Automatic Speech Recognition (ASR) and Natural Language Processing (NLP) applications, they offer a cost-effective solution for training language models. Each dataset is carefully curated to ensure accuracy and reliability.

↓ View speech datasets.

Custom Speech Data

Datasets For Enhanced Speech Technologies

We create high-quality speech datasets, including transcripts, specifically designed for machine learning applications. Our services support technologies aiming to develop or enhance automatic speech recognition (ASR) models using natural language processing (NLP) across select languages and various domains.

Each dataset is customisable based on specific requirements such as dialect, demographics, industry, or other key conditions to ensure optimal model performance. Whether you need data for a particular sector or linguistic group, we provide precisely tailored solutions.

In addition to our curated speech datasets for select languages and industries, we also offer bespoke speech collection projects on request. Our expert team ensures that every dataset meets the highest quality standards, providing the accuracy and linguistic diversity essential for robust ASR and NLP development. Contact us to discuss your specific dataset needs, whether for off-the-shelf solutions or custom projects aligned with your technology’s requirements.

Steps To Order Speech Data

High-Quality Speech Data, Custom-Built For your Needs.

STEP 1

Submit Your Requirements

Use the Custom Speech Request form below to provide details of your speech dataset needs. Our team will review your request and send you a customised job proposal, including pricing and a project timeline, for approval.

STEP 2

Dataset Creation

Once you approve the plan, we begin the process of recording the required speech and producing high-quality transcriptions. Our team ensures the dataset meets your specified criteria, including language, dialect, and industry-specific requirements.

STEP 3

Receive Your Dataset

Upon completion, or at agreed milestones, we deliver your speech dataset—including high-quality recordings and corresponding transcripts—securely and in your preferred format.

Custom Speech Data Form

Request Speech Data For your Needs.

1Personal Details

2Billing Details

3Job Details

4Instructions

Complete your contact details.
Name*
First Last
Email*
Phone
Use full international dialing code, eg 0044 20 999999

Individual or Company?*
Are you submitting this request as an individual or on behalf of a company?
- Individual
- Company
Position
Company Website
Billing Address*
Street Address Address Line 2 City County / State / Region ZIP / Postal Code Country

Sector*
Your service industry.
List Other industry
Language
- English
- Other
English Dialect/s*
List the main dialect/s that you require for your recordings, e.g. "US English", "Australian English", "Scottish English". Be as detailed as you can, e.g. "US English - Texan"
Other Language*
List the language you would like Speech Collection for.
Other Language Dialect/s
List the main dialect/s that you require for your recordings, e.g. "European French", "Congolese French". Be as detailed as you like.
Volume Requirement*
- Once-off
- Ongoing (Monthly)
Volume Once-Off (Audio Hours)
State the ESTIMATED number of audio hours required for Speech Collection.
Please enter a number from 0 to 1000000.
Volume Ongoing (Audio Hours)
State the ESTIMATED MONTHLY average number of audio hours required for Speech Collection.
Number of Months*
You selected "Ongoing". Please indicate the number of months you foresee this requirement, eg 12 (months). At this time we limit the SLA to 48 months.
Please enter a number from 0 to 48.

Start Date
Indicate your preferred date to start if available. NOTE: Depending on the volume and other conditions, we reserve the right to include a lead period of time to set up the required logistical support and quality assurance.
MM slash DD slash YYYY
Details*
Outline your project and list any requirements. You can include a proposed format and codes to be used (or attach an instruction sheet below). Add anything else you think is important for us to consider.
Sample Audio Recording + Transcript File
If you have any original dataset of your required collection, attach a sample audio recording of no longer than 60 seconds and a matching transcript of that audio clip. Please try to keep the attachment under 2MB
Drop files here or

Max. file size: 5 MB, Max. files: 5.
CAPTCHA
Phone
This field is for validation purposes and should be left unchanged.

Frequently Asked Questions

Speech Collection Services

Who uses your Speech Collection service?

Our Speech Collection service is available to clients that want to create or improve existing automatic speech recognition models. Off-the-shelf datasets are available for these purposes, which comprise of unscripted, natural conversations that are conducted by participants recruited, trained, and approved to simulate real-world conversations in common domains. For custom datasets that require specific dialects, languages, domains or conventions, please get in touch to learn more.

Do you specialise in any languages or dialects?

Way With Words has completed Speech Collection projects across a range of English dialects, including Australian, Irish, Scottish, South African and Welsh. With a strong presence in Africa, we have also completed Speech Collection projects in languages such as Afrikaans, isiZulu and seSotho.

Which domains have your Speech Collection services included?

Way With Words has created datasets across many domains, including healthcare, insurance, telecom, finance, retail, fast food, travel, airline, and many more. Custom domains can be commissioned to exact client requirements.

Do you sign Service Level Agreements?

For ongoing work, we prefer to work with an SLA. The SLA sets out a clear timetable that includes an initialisation period to set up the required team and logistics for client work. The SLA also covers terms and conditions related to the work and data privacy. If a client requires ongoing work, over an agreed period, Way With Words also usually provides a dedicated MTP team with management oversight, recruitment, selection, assessment, training processes and any other logistical assistance to aid the bespoke requirement.

Datasets Available for Purchase

Explore Our Ready-to-Use Speech Datasets

High-Quality Speech Data for AI & Machine Learning

Our speech datasets are meticulously planned, collected, annotated, and curated following natural language processing (NLP) best practices.
Designed to support machine learning and speech recognition technologies, our datasets provide unbiased, fully representative speech data with diverse demographic coverage and an optimised gender balance.

Why Choose Our Speech Datasets?

Built for ASR, NLP, and AI model training
Collected from a wide range of demographics
Ensures balanced gender representation
Available for immediate download
Supports benchmarking and accuracy improvements

Dataset Specifications

Hours Available
Age Range
Number of Speakers
Audio Format
Accents

Click below for details ↓

Scottish Accented English Speech Collection

Proven Expertise in Speech Data Collection

Speech Collection Use Cases

Our Speech Collection service has been instrumental in helping clients enhance speech and voice recognition technologies. We have successfully delivered high-quality speech datasets for automatic speech recognition (ASR) and acoustic modelling, ensuring optimal accuracy for machine learning applications.

We have worked across multiple languages and dialects, collecting diverse speech samples to meet the specific requirements of AI-driven speech processing. Below are some of our completed projects:

Afrikaans Call Recording – Afrikaans
Scottish Accented English Speech Collection – English (Scottish Dialect)
UK Accented English Speech Collection – English (UK Dialect)
UK Expats English Speech Collection – English (UK Expats)
Irish Accented English Speech Collection – English (Irish Dialect)
Australian Accented English Speech Collection – English (Australian Dialect)

With extensive experience in curated and custom speech dataset creation, we continue to provide high-quality speech data solutions for ASR, NLP, and voice technology applications.