I spent last Tuesday building a contact tracker for a side project. Opened a browser tab, said five sentences, and had a working table with columns, data, and a filtered view. No spreadsheet, no SQL, no clicking through a setup wizard.
This is Voice Tables, a tool we built at Inithouse. It turns voice commands into structured data. You describe what you need and it builds tables, adds rows, runs queries, and exports results. The entire interaction happens through a microphone.
Here is the exact sequence I walked through, start to finish.
Step 1: open Voice Tables and allow microphone access
Go to voicetables.com and click the microphone icon. Your browser will ask for mic permission. Grant it and you are live. The workspace starts empty, just a text input and a voice button.
One thing to get right early: use a quiet room or a headset. Voice recognition accuracy drops fast in noisy environments. A laptop mic in a coffee shop will produce garbage. A basic headset in a quiet room works well.
Step 2: create a table by describing it
Say something like: "Create a table called Contacts with columns name, email, company, and last contacted."
Voice Tables parses the structure from your sentence. It picks up the table name, infers column types (text for name and company, email for the email field, date for last contacted), and builds the table. The whole thing takes about two seconds.
You do not need to match a specific syntax. "Make me a table for tracking leads with name, phone number, and deal stage" works the same way. The AI figures out the schema from natural language.
Step 3: add data by talking
Once the table exists, adding rows is just describing them: "Add a contact: Marie Novak, marie@example.com, Startup XY, today."
The system maps each value to the right column. If you have a date column, it understands "today," "last Monday," or "three days ago." If you have a number column, it parses numbers from speech without you spelling digits.
A practical tip: pause briefly between values. Saying "Marie Novak comma marie at example dot com" in one breath sometimes confuses the speech-to-text layer. A one-second pause between fields gives cleaner results.
Add a few more rows the same way. I added six contacts in about 90 seconds, which would have taken me longer in a spreadsheet because I would have been tabbing between cells and fixing autocorrect.
Step 4: query your data with a question
This is where it gets interesting. Instead of writing a filter formula or a SQL WHERE clause, you ask a question: "Show me contacts I have not reached out to in more than 30 days."
Voice Tables translates the question into a filter on the last contacted column and returns matching rows. The query runs against your local data, and the results appear as a filtered view of the table.
You can stack queries: "Now sort them by company name" or "Only show contacts from companies that start with S." Each follow-up command refines the view without resetting the previous filter.
Step 5: rename, restructure, adjust
Need to rename a column? Say: "Rename the column last contacted to last interaction."
Want to add a column after the fact? "Add a column called notes, text type." The column appears at the end of the table, and existing rows get an empty value in that field.
Deleting works too: "Remove the phone number column." Voice Tables asks for confirmation on destructive operations, which saves you from accidental data loss when the mic picks up something you did not intend.
Step 6: export what you have
When you are done, say: "Export this table as CSV." The file downloads to your browser's default download folder. You can also say "Export as JSON" if you need structured output for an API or a script.
The export captures the current filtered view, not the entire table. So if you filtered to contacts older than 30 days in step 4 and then export, you get only those rows. Export the full table by clearing filters first ("Show all contacts").
Where this actually saves time
The pitch is not "voice is always better than typing." It is narrower than that. Voice input saves real time in two specific situations:
First, when you are collecting data away from your desk. A fitness coach logging client sessions between appointments. A real estate agent noting property details during a walkthrough. A consultant capturing meeting notes. In all of these cases, the alternative is scribbling on paper and transcribing later, or thumb-typing on a phone.
Second, when the schema is simple and the data entry is repetitive. If you are building a 50-column financial model with nested formulas, use a proper spreadsheet. If you are tracking 6 fields across 30 rows, talking is genuinely faster than typing.
Voice Tables is an Inithouse product. We build tools that solve narrow problems well rather than trying to replace Excel. The workspace also supports docs and an AI chat alongside the tables, but the voice-to-table pipeline is the core of it.
Try it at voicetables.com. Creating a workspace takes about 60 seconds, and the free tier covers more than enough to test whether the voice workflow fits how you actually work.
Top comments (0)