Skip to main content
VerifiedHuman-curatedEnglish → BurmeseParallel corpusNLP / MT / LLM training

EN–MY Professional
Translation Memory Corpus

4.6 million human-verified segment pairs from eight years of professional media localization. Never published online. Zero synthetic data. Built to national Myanmar broadcast standards.

4.6M
Segment pairs
2016–2024
Collection span
0%
Synthetic data
JSONL / Unicode
Format

Data Appraisal

Cryptographically verified. Proof, not promises.

This dataset has passed Aseryx's two-layer cryptographic appraisal. Provenance verified. Information richness scored. No raw data was transmitted at any point in the process.

CompletenessAccuracyUniquenessAuthenticity

Quality Certificate

EN–MY Translation Memory

MAY-2026-EN-MY-001

Verified
0
/ 100
Verified
Completeness
94
Uniqueness
99
Accuracy
91
Timeliness
88
NLP • Parallel corpus

Provenance

Eight years of professional localization workflows.

This corpus was produced for professional media localization between 2016 and 2024. It has never appeared online — no contamination risk for model pre-training or benchmarking. A genuine differentiator against web-scraped alternatives.

01

Outsourced translation

~80 professional freelancers

02

In-house curation

15–25 full-time editors

03

Client review

National broadcast standards applied

04

Corrections & sign-off

3–4 rounds per batch

Many translations produced for national Myanmar TV — adhering to Burmese national language standards, orthography, and broadcasting requirements.

Domain Coverage

8 domains covered.

Medical / clinic talks
Business
Engineering
General conversational
Documentary
Shopping & retail
Relationships
General media localization

Use Cases & Restrictions

Clear terms. Agreed in writing.

Permitted under written license

Research & academic use
Commercial AI/ML training
Machine translation systems
Model evaluation & benchmarking
Internal product development
LLM fine-tuning & RLHF workflows

Restricted unless separately agreed

Raw dataset redistribution
Resale or sublicensing
Public release without agreement
Third-party transfer

Licensing

Three models available.

Default

Non-exclusive

Multiple buyers may license simultaneously. Standard for commercial AI training and research.

Evaluation subset

100K–1M segment samples under separate terms for pre-purchase validation.

Time-based exclusivity

2–3 year exclusive terms available. Scope, use case, and pricing negotiated per deal.

Access control: seller approves each buyer individuallyPlatform fee: 20% on Aseryx-sourced transactions

Request Access

One appraisal away from knowing what this data is worth.

Submit your organization, intended use case, and segment volume. The data owner reviews and responds to each request individually.