The best speech recognition software of 2026

Wispr Flow works anywhere you can type, with no plugins, on Mac, Windows, iPhone and Android. You talk, and Flow removes the ums and filler words, adds punctuation and formats the text before it lands in your email, chat window or document. That cleanup is the main reason to pay for it over the free dictation built into your computer. Flow learns names and unusual terms on its own, or you can add them to a dictionary, and you can save snippets that expand when you say a shortcut. It handles more than 100 languages, and it works best when you pick the language rather than leaving it on auto-detect.

Prices per user on November 4, 2026:

• Free: a limited number of words each week, no card needed.
• Pro, $12 a month billed yearly or $15 monthly: no word limit.
• Growth, $18 a month billed yearly or $23 monthly: adds basic admin controls and organization-wide HIPAA settings.
• Enterprise: custom, billed yearly.

Wispr says it is SOC 2 Type II and ISO 27001 certified, lets you opt out of model training and delete your data, and offers HIPAA-ready dictation with a signed BAA. Its meeting notetaker runs on Mac and Windows only and is switched off on BAA accounts.

Choose it if you write a lot of messages and want cleaner text than built-in dictation gives you. Skip it if you need dictation to work offline, or if you dictate only now and then; the free tools below will do.

Dragon comes from Nuance, which Microsoft owns, and it is built for people who dictate for a living: alongside Dragon Professional there are separate editions for legal work (Dragon Legal) and police reports (Dragon Law Enforcement). The current desktop version, Dragon Professional v16, runs on Windows 11 and Windows 10. You can add custom words, create voice commands that insert boilerplate text or a signature, and record macros that run several steps at once. Dragon can also transcribe recordings from a single speaker, and its Auto Transcribe Folder Agent works through a folder of audio files in a batch.

For people who don't work at one desk, Dragon Professional Anywhere and Dragon Legal Anywhere run in the cloud, and Dragon Anywhere Mobile brings dictation to iPhone and Android in the US and Canada, with custom words synced across devices.

Nuance's site no longer lists prices for any Dragon product. Volume licensing goes through a contact form, and the customer service line is (800) 654 1187. The page mentions no free trial, and there is no Mac version.

Choose it if you dictate long documents on Windows, use a lot of specialist terms and want macros you can trigger by voice. Skip it if you use a Mac, or if your dictation is mostly short messages; Wispr Flow is cheaper to try and works everywhere.

Otter.ai is for conversations rather than dictation. Its meeting agent joins Zoom, Microsoft Teams and Google Meet calls, writes a live transcript with speaker names, and sends a summary with decisions and action items afterwards. If you'd rather not have a bot in the call, you can record from the desktop app, the Chrome extension or your phone. You can then ask Otter questions across all your meetings and draft follow-ups from them.

Prices per user on November 4, 2026:

• Basic, free: 300 minutes a month, up to 30 minutes per conversation, 3 file imports in total and your 25 most recent conversations.
• Pro, $8.33 a month billed yearly or $16.99 monthly: 1,200 minutes a month, up to 90 minutes per conversation and 10 imports a month.
• Business, $19.99 a month billed yearly or $30 monthly: unlimited meeting transcription, 4 hours per conversation and a 7-day free trial.
• Enterprise: custom.

Transcription covers English, Spanish, French, German, Japanese and Chinese. Pro connects to Zapier, Salesforce and HubSpot, and Otter's notes sync to Slack, Notion and Google Docs.

Choose it if you sit in a lot of calls and want searchable notes without typing. Skip it if you mostly want to write by voice, or if your meetings run past 30 minutes and you won't pay, since the free plan stops there.

Every Mac has dictation built in. Turn it on in System Settings under Keyboard, put the cursor where you want text, and press the microphone key or your dictation shortcut. It works in any app that takes text, adds commas, periods and question marks for you in supported languages, and lets you say emoji, punctuation, "new line" or "new paragraph" by name. There's no length limit; it stops after 30 seconds of silence.

On macOS 27, Macs with an M3 chip or later and at least 12GB of memory run an on-device English model that improves spelling, punctuation and capitalization. Keyboard settings show whether your dictation is processed on your Mac or sent to Apple's servers. If you need to control the whole computer by voice, turn on Voice Control instead, which takes over dictation as well.

Choose it if you want to try dictation before paying for anything. Skip it if you need filler words removed or specialist vocabulary learned; Wispr Flow does both.

Windows 11 has voice typing built in: press the Windows logo key and H in any text box, wait for "Listening...", and talk. You can say punctuation by name, say "new line" or "new paragraph", and switch on automatic punctuation from the gear menu. Microsoft lists about 50 languages for voice typing on Windows 11, while the older Windows 10 dictation supports fewer and takes commands in US English only.

To dictate in another language, add it under Settings, Time & language, Language & region, then switch with the Windows key and Spacebar. Voice typing sends your speech to Microsoft's Azure speech service, so it needs an internet connection. For hands-free control of the whole computer, Windows has a separate feature called Voice Access.

Choose it if you're on Windows and want to dictate an email or a document now, for free. Skip it if you work offline or dictate all day with specialist terms; Dragon Professional is the Windows tool for that.

Google Docs has voice typing under Tools, and it works in the latest Chrome, Edge or Safari. Click the microphone, talk, and click again to stop. What sets it apart from the other free options is the editing: in English you can select a word or paragraph, make text bold or a heading, add a comment or footnote, insert a table and move the cursor, all by voice. Say "stop listening" to pause and "undo" to reverse a mistake.

Voice typing supports many languages and accents, but punctuation and editing commands work only in English, and your account and document languages both need to be English. On a work or school account, your administrator may have switched it off.

Choose it if your writing happens in Google Docs and you want to edit without touching the keyboard. Skip it if you need dictation in other apps; it works only inside Docs.

MacWhisper runs transcription models on your Mac, so recordings of interviews, patient notes or client calls never have to leave the machine. Drag in an mp3, wav, m4a or mp4 file and it writes a transcript, with batch processing and speaker recognition, then exports to .srt, .vtt, .txt or .docx for subtitles or editing. If you'd rather use a cloud model, it also connects to OpenAI, Anthropic and Google, or to local models through Ollama and LM Studio.

The free version stays free. Pro is a one-time purchase with lifetime updates and adds system-wide dictation, meeting recording, watched folders and workflow integrations; the site showed a euro price to us, so we couldn't confirm the US price. Good Snooze, Jordi Bruin's studio in the Netherlands, makes it, and there's an App Store version and an iPhone and iPad version as well.

Choose it if you transcribe sensitive recordings, or you'd rather pay once than every month. Skip it if you're on Windows, or if you mostly want live dictation in other apps and don't want to pay for Pro to get it.

Speechnotes is a dictation notepad that opens in Chrome or Edge with no download and no account. You talk, say punctuation and formatting commands, and it capitalizes for you; your notes are saved in the browser's local storage on that device, so they're still there when you come back. A Chrome extension adds voice typing to forms and text boxes on other sites, and there's an Android app.

The notepad is free with Google ads, or $1.90 a month for Premium without ads. Speechnotes also transcribes audio files for $0.10 a minute with no subscription, with English speaker labels, timestamps and .srt caption export, plus an API and Zapier for automating it.

Choose it if you want a free, quick place to dictate a draft on a shared or school computer. Skip it if ads bother you and you won't pay $1.90, or if you need your notes on more than one device, since they live in one browser.

Deepgram is an API, so it suits a developer building transcription into a product, a call-center tool or a voice agent rather than someone who wants to dictate an email. Its Nova-3 model handles both recorded files and live audio, Nova models cover more than 60 languages, and billing is by the second with no rounding up to the minute.

Pay-as-you-go prices on November 4, 2026:

• Nova-3, one language: $0.0043 a minute for recorded audio, $0.0077 a minute streaming.
• Nova-3, multilingual: $0.0052 a minute recorded, $0.0092 streaming.
• Whisper Large: $0.0048 a minute recorded.

The Growth plan lowers those rates, to $0.0036 a minute for one-language recorded audio, for example. New accounts get $200 of credit that doesn't expire, with no card needed.

AssemblyAI is the closest rival. It gives new accounts $50 of credit and charges $0.15 an hour for its Universal-2 model on recorded audio, with add-ons such as speaker labels and a medical mode priced by the hour.

Choose Deepgram if you need fast streaming transcription inside your own software. Skip it if you can't write code; Otter.ai or MacWhisper give you transcripts without one.