Does Immersive Reader read tables?
No. Microsoft states it directly in the Immersive Reader SDK reference: “Images and tables are currently not supported.” A table passed to Immersive Reader loses the structure that makes it readable.
The documentation says so explicitly
From the Immersive Reader SDK reference: “Unsupported tags are rendered comparably. Images and tables are currently not supported.”
This is unusually clear for a product limitation. It is not a bug or an edge case — table support is simply not in the feature.
Source: the Immersive Reader SDK reference.
Why a flattened table is worse than no table
A table carries meaning in two directions. A cell means something because of its row and its column. Remove that and you get a sequence of values with no way to tell which heading each belongs to.
Read aloud, a flattened timetable becomes a stream of times and room numbers. A flattened set of results becomes numbers with no measures attached. The information is technically present and practically unusable, which is harder to detect than an outright failure.
What preserves table structure
Reading a table properly means announcing the row and column a value sits in, and letting you ask about a specific one rather than listening to the whole grid.
Darvy keeps tables in their rows and columns, including in documents nobody made accessible, and answers questions about a particular row instead of reading everything. See what Darvy can do.
Related questions
Does Immersive Reader support tables?
No. The Immersive Reader SDK reference states: “Images and tables are currently not supported.”
Does Immersive Reader describe images?
No. The same sentence in the SDK reference covers images as well as tables.
Why does a table read badly with a screen reader?
Because a cell's meaning depends on both its row and its column. If that relationship is lost, the values are read as a flat sequence with no headings attached.
What reads a table with its structure intact?
Darvy keeps row and column relationships in documents, including inaccessible ones, and can answer a question about one row rather than reading the whole table.
More comparisons and guides are collected in the Darvy resources hub.