Government Of India
A-
A
A+
NE-SpeechEmbed
NE-SpeechEmbed is a multilingual speech-text embedding model by MWire Labs for Northeast Indian languages. The model supports semantic speech search, cross-modal retrieval, and audio-text embeddings across Khasi, Garo, Mizo, Nagamese, Kokborok, Assamese, Wancho, and Chakma.
About Model
NE-SpeechEmbed is the first multilingual speech-text retrieval and embedding model developed for Northeast Indian languages by MWire Labs. The model is trained on 73,476 speech-text pairs collected from proprietary MWire Labs corpora and the Vaani dataset. The architecture uses a fine-tuned Whisper-medium speech encoder and an XLM-Roberta-base text encoder aligned through contrastive learning using InfoNCE loss. Both encoders are projected into a shared 768-dimensional embedding space for speech-text similarity and retrieval tasks. Supported languages include Khasi, Garo, Mizo, Nagamese, Kokborok, Assamese, Wancho, and Chakma across Austroasiatic, Tibeto-Burman, and Indo-Aryan language families. NE-SpeechEmbed supports: * Semantic speech search * Speech-to-text retrieval * Audio-text similarity * Cross-modal embeddings * Low-resource multilingual speech applications The model establishes a foundational benchmark for multilingual speech embeddings in Northeast India and consistently outperforms CLAP zero-shot retrieval baselines across all supported languages. Developed and released by MWire Labs under CC-BY-4.0.
NE-SpeechEmbed
Metadata
Attribution 4.0 International (CC BY- 4.0)
MWirelabs
Multimodal Language Model
PyTorch
Open
Sector Agnostic
11/06/26 19:17:04
0
Activity Overview
0
0 
0 
12
Tags
- low-resource-languages
- speech-embeddings
- speech-text-retrieval
- audio-text-retrieval
- multilingual
- speech-language-model
- mwire-labs
- speech
- embeddings
- northeast-india
- low-resource
- retrieval
- whisper
- XLM-RoBERTa
License Control
Attribution 4.0 International (CC BY- 4.0)
Related Datasets
Updated 1 year(s) ago
VAANI: Multi-modal, Multi-lingual Dataset
- VAANI is a multi-modal, multi-lingual dataset designed to represent the rich linguistic diversity of India. It currently includes data from two phases—Phase 1 (80 districts) and Phase 2 (40 districts)—spanning a total of ~21,500 hours of spontaneous, image-prompted speech collected from more than 110K speakers across 120 districts, describing 210K images in 86 languages. From this, 835 hours of transcribed audio data is available, distributed nearly evenly across all 120 districts.
Bengali
Gujarati
Kannada
Nepali
Punjabi
Telugu
Urdu
Sindhi
English
Tamil
Low-Resource Languages
Marathi
Malayalam
Odia
speech transcription
spontaneous speech
Sangtam
Ao
Halbi
Malvi
Sadri
Chhattisgarhi
Bhili
Malvani
multilingual corpus
Sumi
Bagheli
Khorth
Nyishi
multimodal dataset
Shekhawati
Bagri
Mewati
Meitei
86 Indian languages
Surgujia
audio-visual dataset
Garo
MagadhiMagahi
TTS training
Wancho
Awadhi
Galo
Oriya
speech + image + text
Tenyidie
regional dialects
Rajbanshi
Nagamese
manual transcription
Thethi
LLM speech integration
KhorthKhotta
Konkani
quality evaluated
speaker identification
diverse demographics
telemedicine AI applications
Marwadi
speaker diversity
Tulu
Assamese
Harauti
Rajasthani
language identification
120+ districts
Tagin
Marwari
Gondi
Bajjika
22 Indian states
Surjapuri
dialect diversity
Kokborok
image-prompted data
Santali
speech enhancement
Khariboli
Rengma
geo-centric data collection
Hajong
Hindi
ASR training
Wagdi
Bhatri
Dorli
multi-modal language resources
reallife recording environments
Mewari
linguistic diversity
Nimadi
Khandeshi
data for conversational AI
benchmarking dataset
Lotha
Kurmali
Bhojpuri
NissiDafla
Angika
Lambani
Magahi
Jaipuri
Bihari
Chakhesang
Magadhi
Angami
Chakma
Duruwa
Bearybashe
Khortha
Kumaoni
Kurukh
Garhwali
Maithili
Agariya
Bundeli
large-scale speech corpus
- 1
- 163
- 0
1,884
More Models from MWire Labs
Northeast STT Multilingual Speech to Text Model
A multilingual Speech-to-Text (STT) model for eight Northeast Indian languages, fine-tuned from Whisper Medium using over 150,000 speech-text pairs from public and institutional datasets. The model expands speech recognition support for low-resource indigenous languages, including Khasi, Garo, Mizo, Kokborok, Nagamese, Assamese, Chakma, and Wancho.
Northeast India Languages
northeast-india
low-resource
Automatic Speech Recognition
Multilingual
Speech to Text
Multilingual speech
Speech processing
whisper
- 0
- 0
- 0
- 16
Updated 4 day(s) ago
NE-SpeechEmbed
NE-SpeechEmbed is a multilingual speech-text embedding model by MWire Labs for Northeast Indian languages. The model supports semantic speech search, cross-modal retrieval, and audio-text embeddings across Khasi, Garo, Mizo, Nagamese, Kokborok, Assamese, Wancho, and Chakma.
speech-embeddings
speech-text-retrieval
audio-text-retrieval
multilingual
speech-language-model
mwire-labs
speech
XLM-RoBERTa
low-resource
retrieval
northeast-india
low-resource-languages
whisper
embeddings
- 0
- 0
- 0
- 13
Updated 7 day(s) ago
NE-Embed
NE-Embed is a multilingual text embedding model for Northeast Indian languages, enabling semantic search, retrieval, and RAG across 10 languages including Khasi, Garo, Meitei, Bodo, Mizo, Assamese, Nyishi, Kokborok, Pnar, and Nagamese. Fine-tuned on LaBSE with 201,738 parallel pairs.
retrieval
Multilingual
rag
embeddings
northeast-india
sentence-transformers
low-resource
- 0
- 0
- 0
- 14
Updated 7 day(s) ago
Mizo OCR - Text Recognition for Mizo Language
OCR model for the Mizo language achieving 90.68% character accuracy on synthetic and curated printed text
Image-to-Text
OCR
low-resource
northeast-india
Mizo
trocr
- 0
- 2
- 0
- 257
Updated 3 month(s) ago
NE-OCR
NE-OCR is a multilingual Optical Character Recognition model developed by MWire Labs to accurately recognize printed text from documents in Northeast Indian languages. The model supports Assamese, Bodo, English, Garo, Hindi, Khasi, Kokborok, Meitei (Bengali script), Meitei (Meitei Mayek script), Mizo, Nagamese, and Nyishi. It is designed to enable reliable digitization of books, newspapers, government records, educational materials, and cultural archives from Northeast India where mainstream OCR
vitstr
Multilingual OCR
Northeast India OCR
Printed Text Recognition
OCR
Garo
Kokborok
Nyishi
Meitei
northeast-india
Nagamese
khasi
Mizo
Optical Character Recognition
BODO
doctr
- 0
- 8
- 0
- 186
Updated 3 month(s) ago
Nagamese Speech-to-Text
Automatic Speech Recognition (ASR) model for Nagamese speech, designed to transcribe spoken Nagamese into text for real-world usage.
Automatic Speech Recognition
low-resource-language
whisper
Nagamese
Speech Recognition
ASR
- 0
- 3
- 0
- 186
Updated 3 month(s) ago
Garo OCR - Text Recognition for Garo
OCR model for the Garo language achieving 93.13% character accuracy.
Garo
northeast-india
florence-2
Image-to-Text
OCR
- 0
- 0
- 0
- 249
Updated 3 month(s) ago
Northeast Language Identification
NE-LID is a fast and accurate language identification model for Northeast Indian languages using character level features. It is designed for low resource and script diverse text and achieves high accuracy on short sentences.
low-resource
Multilingual
language identification
fastText
fasttext
MWire Labs
northeast-india
- 1
- 13
- 0
- 813
Updated 6 month(s) ago
NortheastNER
NortheastNER is a token classification model built on XLM-RoBERTa and fine-tuned on ~25k sentences from gazetteers, news, and cultural texts across Northeast India. It detects region-specific entities, places, tribes, festivals, tourist sites, flora, fauna, and experimental local names; ideal for low-resource NER, regional search, cultural analytics, and knowledge graph applications.
northeast-india
low-resource
XLM-RoBERTa
NER
Token Classification
Conservation
Northeast India
Meghalaya
- 0
- 14
- 0
- 350
Updated 7 month(s) ago
Kren-M
Northeast India's first AI language model. Kren-M is a 2.6B parameter bilingual model for Khasi-English, built on Gemma-2-2B. Features Kren-NE custom tokenizer covering 7 NE languages (Khasi, Garo, Mizo, Assamese, Manipuri, Nagamese, Nyishi) with 35.7% efficiency gain. Trained on 5.43M Khasi sentences. Capabilities: bidirectional translation, natural conversation, cultural context. Designed for language preservation across Northeast India
Garo
low-resource
Instruction-Tuning
Indian Languages
continued-pretraining
bilingual
Kren-M
Northeast India Languages
Foundational model
Tokenizer
Northeast India
khasi
northeast-india
- 0
- 34
- 0
- 963
Updated 7 month(s) ago
© 2026 - Copyright AIKosh. All rights reserved.