Automotive / In-Cabin AIeBanma (Banma Network)2026-08-01

eBanma: Audio, Transcripts and Embeddings in One Queryable Engine

An in-cabin assistant improves only as fast as its training data. eBanma aggregates the fleet's daily voice and language volume onto MOI, standardised on a common event schema, with audio, transcripts, embeddings and labels in one engine — so a training corpus is queried rather than hunted for.

eBanma (Banma Network)

eBanma (Banma Network) builds intelligent vehicle operating systems and connected-car solutions, founded in November 2015 with Alibaba Group as its largest shareholder and SAIC as its innovation source. Built on its self-developed AliOS, its cockpit solutions ship with more than ten auto brands including SAIC, FAW and Volkswagen Group.

One engine
Audio/transcript/vector/label
Common schema
Fleet events standardised
SQL + vector
Dual retrieval
With context
Vehicle and session

The challenge

An in-cabin assistant improves only as fast as its training data. Vehicles on the road generate an enormous daily volume of voice interactions and the language data around them-utterances, transcripts, recognized intents, dialogue context, correction and fallback events-across models, regions, accents and cabin conditions. Making that stream useful for deep-learning training means landing it at scale, keeping it labelled and queryable rather than as loose files on storage, assembling training sets that target specific weaknesses, and doing all of it under the privacy controls that in-cabin audio demands.

The solution

eBanma aggregates voice and language data onto one OmniFabric platform. Data streams in continuously from the fleet, is standardized on a common event schema, and is stored alongside its structured context-vehicle model, region, cabin state, dialogue session-with transcripts, embeddings and labels held in the same engine. Teams assemble training sets by query, in SQL and by vector similarity: the utterances where the assistant fell back, the accents a model handles badly, the sessions around one intent. Versioned datasets feed deep-learning training and evaluation, and results are written back as labels so the next iteration targets what actually failed.

The outcome

Audio, transcripts, embeddings and labels sit in one engine with their vehicle and session context, so the training corpus is searched rather than hunted through directories.

A new experiment begins as a query-this intent, this region, these fallback cases-instead of a data-collection project, so iteration speed is set by the model team rather than the pipeline.

Vehicle model, region, cabin state and dialogue session stay attached to every utterance, which is what makes it possible to isolate where a model actually underperforms.

Solution Architecture

Data sources
  • In-Cabin Voice Interactions from the Fleet
  • Transcripts, Intents & Dialogue Context
  • Fallback, Correction & Error Events
  • Vehicle, Region & Cabin Metadata
MatrixOne Intelligence
  • High-Volume Continuous Ingest from the Fleet
  • Audio, Transcript, Embedding and Label in One Engine
  • SQL and Vector Query to Assemble Training Sets
  • Versioned Datasets for Training and Evaluation
  • Privacy Controls and Access Governance on Cabin Data
Business applications
  • Queryable Voice & Language Corpus
  • Targeted Training Sets on Demand
  • Versioned Datasets with Lineage
  • Evaluation Results Fed Back as Labels

Related solutions and products