How much money have dictation and voice AI startups raised?

Money in this field is lopsided. A handful of deals are worth hundreds of millions or more, while most rounds are far smaller. The largest acquisition on record is Microsoft's purchase of Nuance for $19.7 billion million US dollars ¹. The largest venture round is ElevenLabs' Series D of 500 million US dollars ². If you want the typical round for a kind of product, use the medians by segment below, not the headlines.

Our dataset holds 50 data points ³, each a disclosed amount traced to one of 46 sources ⁴. It covers rounds, acquisitions and IPO raises of companies whose main product is speech recognition, dictation, transcription, meeting notes, a clinical scribe or voice AI.

The biggest deals are few and sit at the top of the scale

The biggest figures are not all the same kind of money. Nuance was an acquisition, and the price includes the company's net debt. The iFlytek figure of 3,999,999,986.84 million US dollars ⁵ is a share placement by a listed company, converted from yuan. ElevenLabs, Wispr Flow and Tandem Health are venture rounds.

{{data:voice-ai-funding||limit=8}}

Two recent venture rounds show how far apart the segments are. Wispr Flow, a dictation app, announced 280 million US dollars in a Series B ⁶. Tandem Health, a clinical scribe, announced $100 million US dollars ⁷.

The typical round differs a lot by segment

Medians are more honest than totals here, because one deal can swamp a segment's sum. These are the medians in the data:

The segments rest on very different numbers of rows: 14 for meeting notes, 11 for voice agents and voice AI, 8 for speech-to-text APIs, 6 for medical scribes, and 3 each for dictation apps and transcription services. The dataset has no rows for "Speech models" or "Other". A median of 3 rows moves a lot when one more round is added.

Meeting notes and speech-to-text APIs

Meeting notes is the busiest segment. Granola's Series C was 125 million US dollars ¹⁴, and Read AI's Series B was 50 million US dollars ¹⁵. Among speech-to-text APIs, Deepgram's Series C was 130 million US dollars ¹⁶, and an earlier AssemblyAI Series C was 50 million US dollars ¹⁷.

Dictation apps and transcription services

Dictation apps have only 3 rows. Wispr Flow's Series A was 30 million US dollars ¹⁸, and its later extension was 25 million US dollars ¹⁹. Transcription services raise the least: Amberscript's Series A was 10 million US dollars ²⁰.

What these numbers do and do not measure

The figures are only as complete as public disclosure. Several limits matter:

• Undisclosed rounds are missing. A company that kept its amount private has no row, so the data leans toward companies that like to announce.
• Valuations are excluded. Only money raised or paid counts. Headline valuations are a different number and not comparable.
• Some figures are cumulative or approximate. Gradium's 100 million US dollars is a seed total after an extension, not the size of that extension ²¹. The iFlytek amount is converted from yuan at a rate read in a news report, so it is approximate. Noota's amount was stated in euros and converted.
• Round labels are uneven. Some companies do not name their stage, so those rounds are filed as "Growth or unlabeled round". Tandem Health's rounds are filed that way.
• Some rows come from news reports. For example, the Wispr Flow Series A source reports a round being raised, not one that had closed.
• One date is disputed. Smallest.ai's Series A of 13 million US dollars ²² is dated by the company's page, while press reports give an earlier date for the same round.
• Extremes skew ranges. The range for medical scribes runs from $9.5 million US dollars ²³ up to the Nuance deal, so read its median, not its average.

Funding also says nothing about quality or accuracy. A large round shows investor appetite, not that a product dictates better.

What this means when you choose a dictation tool

Funding tells you about a company's staying power and direction, not about how well it recognizes your voice. A well-funded company can run for years, but it may also pivot toward business customers or voice agents rather than the single-user dictation you want. Smaller, quieter companies in transcription can be solid for one job and still raise little.

Use funding as one signal among several. Check what a tool does with your audio, whether it works offline, how it handles your language and terminology, and what it costs. Our guide to AI dictation apps walks through those tradeoffs, and the market comparison puts products side by side. Free built-in tools are often enough for casual use, so a large funding round is no reason to switch.

For the underlying rows, quotes and sources, see the full dataset. For the wider picture of dictation on your devices, start with our overview of dictation software.