Is voice recognition the same as speaker recognition?

No. Speech recognition works out which words were said. Speaker recognition works out who is speaking. The IEEE defines speaker recognition as identifying or verifying the identity of an individual from characteristics of their voice ¹. The two can use similar acoustic measurements, but they answer different questions. The confusion comes from the name: "voice recognition" is used for both. For the dictation side, see our overview of dictation software and what dictation software is.

Speech recognition answers what you said

Dictation software is a speech recognition product. Its job is to turn your words into text in the app where your cursor is ². It does not need to know who you are. Two people saying the same sentence should produce the same text, which is the goal of the recognizer. The question on what voice recognition software is used for lists the other uses.

Speaker recognition answers who is speaking

Speaker recognition has two tasks. In speaker verification, the system accepts or rejects a claimed identity ³. In speaker identification, no identity is claimed, and the system searches a set of known speakers for the closest match ⁴. Verification checks whether a claimed person is the one speaking. Identification has no claim to check against, so it searches a whole set of voices.

Dictation profiles adapt to your speech, not your identity

Dictation products often ask you to train them. Nuance's help for Dragon for Mac version 6 says Voice Training helps Dragon learn how you speak, and asks you to read aloud a story ⁵. That adapts the recognizer to your voice and vocabulary. It is not speaker identification, although the word "profile" appears in both. Nuance defines a Dragon user profile as a set of files that Dragon uses to process your speech, which stores information such as the type of microphone you use and your custom words ⁶. The definition lists equipment and vocabulary. It does not describe a check of who you are. The glossary entry on voice profiles explains the term in dictation products.

How to tell which one a product does

Read the product's description of what it stores. A dictation product that asks you to read a passage so that it can learn your speech is adapting its recognizer. A product that checks who is speaking, or that says it matches a voice to a person, is doing speaker recognition. The difference shows in the settings: a dictation profile stores details such as your microphone and custom words ⁶, while a speaker check accepts or rejects a claimed identity ³.

Why the difference matters for privacy

Speaker recognition is a biometric method, and biometric data is treated with care. The FTC's amended children's privacy rule adds biometric identifiers that can be used to recognize a person to the definition of personal information ⁷. A dictation product that only transcribes your words is a different privacy question from one that stores a voice signature. Spoofing is also a concern. The IEEE notes that speaker recognition must contend with spoofing attacks, which can include synthesizing speech with text-to-speech tools ⁸. The question on speech to text and text to speech explains that reverse direction. Read the privacy policy of any product to see which of these it does, and whether audio or transcripts are kept.