5 min read
How many good ideas, client requests, and price quotes are sitting in your phone right now as voice memos and voicemails you will never listen to again?
If the honest answer is “a lot”, speech-to-text is the AI skill worth learning first. It is not flashy, it is cheap or free, and it turns the way you already work (talking while driving, walking the dog, or standing on a job site) into notes you can search, copy, and act on. This guide starts from zero.
What speech-to-text actually means
Speech-to-text, also called transcription, is software that listens to spoken words and writes them down. Older versions needed you to speak slowly and clearly into a headset. Modern AI versions handle normal conversation, background noise, accents, and more than one speaker.
There are three flavors you will run into:
- Dictation: you talk, and words appear on screen as you go. Good for writing emails or notes without typing.
- Recording transcription: you upload or record audio, and you get the full text afterward. Good for voice memos, voicemails, and interviews.
- Meeting transcription: a tool listens to a live call and produces a transcript plus a summary.
Two terms worth knowing, since you will see them on product pages. Diarization means the tool can tell who said what, labeling speakers separately. Context biasing means you can give it a list of unusual words, like your product names or trade jargon, so it spells them correctly.
Why this matters for a one-person business
When you work alone, you are the salesperson, the project manager, and the note taker. Speech-to-text helps with all three:
- Capture ideas the moment you have them, instead of hoping you remember them later.
- Keep records of client conversations, so you can check exactly what was agreed.
- Turn talk into finished work: a rambling voice memo becomes a tidy quote, a blog draft, or a to-do list once you paste the transcript into an AI assistant.
- Work across languages, since many tools now transcribe dozens of languages, which helps if your customers do not all speak English.
The real power comes from pairing the two steps. Transcription gets the words out of your head; an AI assistant shapes them into something useful. We used this exact pairing in our walkthrough on turning a discovery call into a proposal in 20 minutes.
You probably already own a transcription tool
Before you pay for anything, check what is already on your devices:
- Your phone’s keyboard has a microphone button on both iPhone and Android. Tap it in any app and talk.
- Google Docs voice typing works in the Chrome browser for free. Google’s voice typing help page shows how to switch it on from the Tools menu.
- Your phone’s voice recorder on many recent models transcribes recordings automatically.
- AI assistants like ChatGPT, Claude, and Gemini accept voice input in their mobile apps, so you can talk your request instead of typing it.
For meetings, there are dedicated notetakers. If you want one that keeps audio on your own device, our review of Google AI Edge Foresight covers a free option.
What it costs when you outgrow the free tools
Transcription has become one of the cheapest things AI does. As one example, Mistral announced its Voxtral 3 speech models on October 20, saying its smaller transcription model costs $0.0035 per minute through its developer API and covers more than 50 languages. At that rate, an hour of audio costs about 21 cents.
You do not need to touch a developer API to benefit. Prices like that trickle down into the apps you already use, which is why transcription is now bundled into so many tools at little or no extra cost. When you do compare paid apps, look at minutes per month, supported languages, speaker labeling, and whether you can add custom vocabulary.
Your first step: the 10-minute voice memo test
Here is a gentle way to start this week:
- Pick one recurring task you usually type, such as writing a quote, a weekly update, or a social post.
- Record yourself talking it through for two or three minutes, as if explaining it to a friend. Do not worry about “ums”.
- Get a transcript using your phone recorder, Google Docs voice typing, or your AI assistant’s voice mode.
- Paste the transcript into your AI assistant with an instruction like: “Turn this into a short, friendly quote for a client. Keep my prices exactly as I said them.”
- Check every number and name before sending. Transcription mishears things like “fifteen” and “fifty”.
If that saved you time, repeat it for a second task next week. Small habits beat big setups.
A few cautions to keep in mind
- Ask before you record anyone. Many U.S. states and other countries require everyone on a call to agree to recording. Saying “Mind if I record this so I get the details right?” covers you and builds trust.
- Know where the audio goes. Some tools process audio on your device; others send it to the cloud. For sensitive client calls, prefer on-device options or check the tool’s data settings.
- Proofread names and numbers. Accuracy is high but never perfect, especially with accents, crosstalk, and industry jargon.
- Translation is a separate step. A transcript in Spanish turned into English by AI is a good first draft, but as we argued in our piece on AI translation, have important text reviewed before customers see it.
Where to go from here
Once voice memos feel natural, you can try transcribing client calls, recording quick video walkthroughs, or dictating whole articles. But start with one memo and one task. The best AI habit is the one you still use a month from now.
What is the first thing you will try dictating instead of typing? Let us know in the comments.
