Speech-to-Text Tools: How They Work and What to Choose
What speech-to-text does and why the right tool matters
Speech-to-text converts what you say into written words on a screen or document. The tool listens through a microphone, processes your voice, and types out the text in real time or after you finish speaking. Different tools work in different places — some built into your phone or computer, some as standalone apps, some embedded in specific programs like email or word processors.
The choice matters because accuracy varies, setup time varies, and what works for one person's situation may not work for another. A tool that works beautifully for taking notes might be clumsy for writing emails. One that requires constant internet might drain your phone battery. Another might misunderstand accents or medical terms you use every day. This guide walks you through the main options so you can match the tool to what you actually do.
Key Takeaways
- Built-in tools like Windows Dictation and Apple Dictation are free and require no setup, but they work only in certain apps and have lower accuracy than paid options.
- Google Docs voice typing and Microsoft Word dictation are free within those programs and work well for longer writing, but you must be online and in that specific app.
- Specialized apps like Dragon NaturallySpeaking and Otter offer higher accuracy and work across more programs, but they cost money and may require training the tool to your voice.
- Real-time captioning tools like Live Captions and Otter Live Transcription turn speech into text as it happens, useful for meetings and conversations, though accuracy depends on audio quality.
- Accuracy improves when you speak clearly, use a decent microphone, and choose a tool designed for your specific task — note-taking, document writing, or live conversation.
Built-in tools that come with your device
Every major operating system includes a basic speech-to-text tool at no cost. On Windows, it is called Dictation and lives in Settings under Ease of Access. On Mac, it is Dictation in System Preferences. On iPhone and iPad, it is the microphone button on the keyboard. On Android, it is Google Voice Typing, accessed through the keyboard.
These tools work only in certain apps — usually email, messaging, notes, and web browsers — not across your whole computer. They send your voice to the company's servers (Microsoft, Apple, or Google) to process, so you need internet. Accuracy is decent for everyday speech but lower than paid tools, especially with accents, background noise, or specialized vocabulary. Setup takes seconds: you turn the feature on and start speaking. No training needed.
Built-in tools are the right choice if you dictate occasionally, want zero cost and setup, and do not mind the accuracy trade-off. They are not the right choice if you need to dictate into many different programs or if accuracy matters more than convenience.
Free tools within Google Docs and Microsoft Word
If you write in Google Docs, you can use Voice Typing at no cost. Open a document, go to Tools, and select Voice Typing. A microphone icon appears on the left side. Click it and speak. Google Docs transcribes as you talk and lets you edit afterward. It works on any device with a browser and internet connection.
Microsoft Word has a similar feature called Dictate, available in Word online and the desktop app (depending on your subscription level). Open a document, go to the Home tab, and select Dictate. Like Google Docs, it transcribes in real time and works across devices if you use OneDrive to sync.
Both tools are accurate for general writing and handle punctuation reasonably well if you say "period" or "comma" aloud. The main limitation is that they work only inside that one program — you cannot use them in email, messaging, or other apps. They also require internet. Choose these if you do most of your writing in Google Docs or Word and want a free, low-friction option.
Specialized apps with higher accuracy
Dragon NaturallySpeaking (Windows and Mac) is the longest-established paid tool. It costs around $200 for a one-time license or $20 per month for a subscription. Dragon works across almost all programs on your computer — email, web browsers, word processors, spreadsheets — and has the highest accuracy of any consumer tool. It learns your voice and vocabulary over time, especially if you train it by reading sample text aloud during setup.
Otter is a newer app available on phone, tablet, and computer. The free version transcribes up to 600 minutes per month and stores transcripts in your account. Paid plans ($10 to $20 per month) offer more minutes and features like speaker identification and searchable transcripts. Otter works across apps on your phone and as a web app on your computer, and it syncs between devices.
Google Recorder (Android and web) records audio and transcribes it afterward, useful if you want to capture a conversation or lecture and read it later. It is free and works offline — transcription happens on your phone, not on Google's servers. The transcript stays on your device unless you share it.
Specialized apps are worth the cost if you dictate frequently, need accuracy above 95%, or want to use speech-to-text across many programs. They are overkill if you dictate rarely or only in one app.
Real-time captioning for meetings and conversations
Live Captions (built into newer Android phones and Chromebooks) transcribes speech happening around you in real time — useful in meetings, lectures, or conversations. You turn it on and it captions what it hears. It works offline on your device.
Otter Live Transcription (free app for phone and web) does the same thing and works across devices. You start a recording and it transcribes as people speak. You can share the transcript with others or save it for later.
Microsoft Teams and Google Meet both include live captioning during video calls. Turn on captions in the meeting settings and they appear in real time for all participants. Accuracy depends on audio quality and how clearly people speak.
Real-time captioning is most useful if you attend meetings, lectures, or conversations where you need to follow along or have a record afterward. Accuracy is lower than tools designed for dictation because the audio is often noisy and people speak at varying speeds. These tools work best in quieter settings with clear speech.
Accuracy, microphones, and what affects performance
All speech-to-text tools perform better when you speak clearly, pause between sentences, and use a decent microphone. A $15 USB microphone or a good pair of earbuds with a built-in mic will give you noticeably better results than your device's built-in microphone, especially in a room with background noise.
Accuracy also depends on the tool itself. Built-in tools and free web-based tools typically reach 85 to 90% accuracy. Specialized apps like Dragon reach 95% or higher, especially after they learn your voice. Real-time captioning tools are usually 80 to 90% accurate because they process speech as it happens rather than after you finish.
Specialized vocabulary — medical terms, technical jargon, brand names — trips up most tools. Dragon and Otter let you add custom words to their dictionary so they learn your field. Built-in tools cannot do this. If you use the same technical terms every day, a specialized tool pays for itself in time saved correcting mistakes.
Background noise is the biggest accuracy killer. Wind, traffic, other people talking, and keyboard clacking all confuse the tool. A quiet room and a microphone that filters background sound make the largest difference in real-world use.
Privacy and where your voice goes
Built-in tools and free web-based tools send your voice to the company's servers — Microsoft, Apple, or Google — to process. They typically delete the audio after transcription but keep the text. Read the privacy policy for the specific tool if you have concerns about what happens to your recordings.
Google Recorder and some versions of Otter process speech on your device itself, not on servers. The transcript stays on your phone unless you choose to share it. This is more private but may be slower on older devices.
Dragon NaturallySpeaking runs entirely on your computer. Your voice never leaves your device. This is the most private option if that matters to you.
If you dictate sensitive information — medical details, financial data, legal matters — choose a tool that processes on your device or read the privacy terms carefully. Some workplaces have policies about which tools you can use with confidential information.
Choosing the right tool for your situation
| Your situation | Best choice | Why |
|---|---|---|
| You dictate occasionally, want zero cost | Built-in Dictation or Voice Typing | Free, no setup, works for basic use |
| You write mostly in Google Docs or Word | Voice Typing or Dictate | Free within the app, good accuracy, no learning curve |
| You dictate frequently across many programs | Dragon NaturallySpeaking | Highest accuracy, works everywhere, learns your voice |
| You want to record and transcribe later | Google Recorder or Otter | Captures audio, transcribes accurately, searchable |
| You need captions in meetings or conversations | Live Captions or Otter Live Transcription | Real-time, works across devices, shareable |
| You use specialized vocabulary daily | Dragon or Otter | Custom dictionaries, learns your field |
| Privacy is your top concern | Dragon or Google Recorder | Processes on your device, not on servers |
Getting started with a new tool
Start with what you already have. If you have a smartphone or computer, you already own a speech-to-text tool. Use it for a week in your normal routine. Notice what works, what frustrates you, and what you wish it could do. This tells you whether you need something more powerful.
If the built-in tool is not accurate enough or does not work in the programs you use, try the free version of a specialized app — Otter's free tier or Google Recorder — before spending money. Most paid tools offer a free trial or free tier so you can test them without commitment.
When you switch to a new tool, give it time to learn. Speak clearly for the first few days. If the tool offers training (Dragon does), spend 10 minutes reading the sample text aloud. This teaches it your accent and speech patterns. Accuracy improves noticeably after training.
Keep a list of words the tool gets wrong and add them to the custom dictionary if the tool supports it. Over a few weeks, accuracy climbs as the tool learns your vocabulary and style.
Frequently Asked Questions
Do I need to be online to use speech-to-text?
It depends on the tool. Built-in tools, Google Docs Voice Typing, and Microsoft Word Dictate all require internet because they send your voice to servers. Dragon NaturallySpeaking, Google Recorder, and some versions of Otter work offline on your device. Check the tool's documentation if internet access is a concern.
Can speech-to-text understand different accents?
Yes, but accuracy varies. Built-in tools and free web tools handle common accents reasonably well but may struggle with less common ones. Specialized tools like Dragon improve significantly after you train them with your voice. If you have a strong accent or speak English as a second language, training the tool makes a real difference.
What if the tool keeps getting the same word wrong?
Most paid tools and some free tools let you add custom words or corrections to a dictionary. Dragon and Otter both have this feature. For built-in tools, you usually have to correct the word each time it appears. If a tool does not support custom dictionaries and you use the same specialized words daily, a tool that does support them will save you time.
Can I use speech-to-text in a noisy environment?
It works, but accuracy drops significantly. A microphone that filters background noise helps more than any software setting. If you dictate in noisy places regularly, invest in a good headset microphone. For occasional use in noise, expect lower accuracy and plan to edit more.
Is speech-to-text faster than typing?
For most people, yes — speaking is faster than typing. But the time saved depends on accuracy. If you spend 10 minutes dictating and 5 minutes correcting mistakes, you have saved time. If you spend 10 minutes dictating and 15 minutes correcting, you have lost time. Accuracy matters more than speed.