Reverie’s NMT is a neural machine translation API that delivers fast, high-quality, and affordable language translations for Indic languages. NMT is a form of translation automation powered by advanced machine learning and artificial intelligence inventions to deliver more accurate and more natural-sounding translation.
The solution will boost human efforts, taking the friction out of the content supply chain, and delivering reckonable value to the business. It localizes the content available on digital platforms like websites, mobile applications, and more in real-time.
You can use this API to translate the source content from English into Indic languages and vice-versa. This document describes the available variables, commands, and interfaces that make up the Reverie’s NMT API. The API Reference document is for Product Managers, Software Developers to learn how to use software libraries’ features. The API is organized around REST. All requests should be made over SSL. All request and response bodies, including errors, are encoded in JSON.
The NMT API has the following features that recognize it as the best in the content localization industry -
The NMT API will automatically detect the language used in your source text if you have missed or unaware of the source language code.
The NMT solution is a continually trained engine which leverages machine learning to combine knowledge of grammar, language structure that always learns from new and expanded datasets to produce more accurate translations for extensive use cases.
The API supports the glossary customization offering high-quality machine translation for Indic language. It will allow you to define how your brand names, character names, and other unique terms get translated using the Customize Terminology feature. This solution is built with the collaboration of linguistic experts, engineers, and domain experts across multiple industries.
Reverie’s NMT fits within the business applications across multiple business verticals, and content flows seamlessly to increase productivity and efficiency. Localize the massive amount of data in a shorter interval with high-quality and consistency. The solution is reliable and requires minimal administrative skill sets to maintain service levels.
Security is one of the primary goals of our solution. The solution can be deployed either on-premises or in a private cloud that restricts data exposure.
The API will successfully identify the context of a word used in the source content and then translates accordingly. This feature will decrease the editorial work as it can recognize the meaning of the word and help the native language users consume the content in the right order.
Language | Language ISO Code |
---|---|
hindi | hi |
marathi | mr |
punjabi | pa |
gujarati | gu |
tamil | ta |
telugu | te |
malayalam | ml |
kannada | kn |
odia | or |
assamese | as |
bengali | bn |
Element | Type | Is Mandatory? | Description |
---|---|---|---|
src_lang |
string | Yes | Source language to convert from. |
tgt_lang |
string | Yes | Target language to convert to. |
Element | Type | Is Mandatory? | Description |
---|---|---|---|
data |
array of content | Yes | List of input text for transliteration |
n_best |
integer | No | Mention the number of translation suggestions the API should return for the input text. Note - By default, the n_best = 1. Note - The maximum number of suggestions you can obtain is 5. |
mask |
boolean | No | The feature to screen the non-dictionary words used in a sentence. In other words, the mask will indicate the words that should not be translated into the target language. Note - By default, the mask = false. Note - To set the mask = true, it is mandatory the src_lang = en (English). |
mask_term |
array of content | No | Determines the Words that are to be masked or to be replaced by another word in a sentence. Note - If the list of terms is provided, then it will be transliterated. Note - If only an object is provided, then only the respective value is masked. Note - On defining values in the mask_term, then automatically the mask is set to true. Example - Replacing a word with another - “mask_terms”: {“Bangalore”: “Bengaluru”} Here, the API will mask the term Bangalore and will replace it with Bengaluru. Masking a term - “mask_terms”: [“Reverie Language Technologies”] Here, the API will mask the term Reverie Language Technologies, if found in the source content, and will transliterate the word. |
segment_after |
number | No | The number of characters after which the content has to be segmented into sentences before translation. Note - By default, segment_after = 0 (Zero). Indicates that API will auto-segment the sentences |
Status Code | Status Type | Message | Description |
---|---|---|---|
403 | Forbidden | Invalid REV-API-KEY or REV-APP-ID | Entered Invalid credentials |
500 | Internal Server Error | Internal Server Error | An error occurred with the API. |
403 | Forbidden | usage exhausted | The provided credits or the character limit is exhausted. |
403 | Forbidden | API key expired | The API key provided to a user is expired. |
403 | Forbidden | unauthorized to use this src/tgt language | The invalid language code is passed, or the user cannot use the defined language code. |
403 | Forbidden | unauthorized to use this API | The REV-APPNAME entered is invalid, or the user is not authorized to use it. |
Sample 1
{
"data": ["Reverie Language Technologies was established in 2009.", "The company head office is located in Bengaluru."],
"n_best": 2,
"mask": true,
"mask_terms" : ["Reverie Language Technologies"],
"segment_after": 0
}
{
"status": "ok",
"result": [
[
"रेवरी लॅग्वेज टेक्नोलॉजीस की स्थापना 2009 में हुई थी।",
"रेवरी लॅग्वेज टेक्नोलॉजीस की स्थापना 2009 में हुई।"
],
[
"कंपनी का मुख्यालय बेंगलुरु में स्थित है।",
"कंपनी का प्रधान कार्यालय बेंगलुरु में स्थित है।"
]
],
"masked": [
"Reverie Language Technologies",
"2009"
],
"took": 0.202,
"mask_failed": [],
"src_lang": "en",
"autoDetect": false
}