Voice Mode in Claude
A beginner-friendly guide to talking with Claude out loud: what voice mode is, how the conversation works behind the scenes, which models and tools it can use, how it works in the car, and how to get the best results from it.
Think about the last time you asked a friend for directions or advice. You did not write them a letter and wait for a reply. You simply spoke, they answered, you asked a follow-up, and the conversation flowed. Speaking is the most natural way humans exchange ideas, and for a long time, talking to a computer meant typing. Voice mode changes that. It lets you speak to Claude and hear Claude speak back, so the exchange feels more like a chat with a knowledgeable friend than a series of typed messages.
This guide explains voice mode from the very beginning. You do not need any technical knowledge. Every new word is explained the first time it appears, and each big idea comes with an everyday comparison. Voice features change often, so the details here reflect what was publicly documented at the time of writing. For anything you plan to rely on, the official Claude help pages on support.claude.com are the best place to confirm the latest.
1What Is Voice Mode?
Let us start with the simplest possible definition and build from there.
Voice mode is a feature that allows complete spoken conversations with Claude. You speak your question out loud, Claude listens, and Claude answers with a spoken voice. You can then respond again, and the back-and-forth continues, just as it would with a person.
Voice mode is a beta feature. In software, beta means a feature is available to use but is still being tested and improved, so it may change or have rough edges. It is documented as available across all Claude plans, including the free one, on the Claude mobile apps for iPhone and Android, on Claude Desktop and on the web. It is described as built to work best from a phone, which makes sense, since phones have a microphone and speaker always at hand.
Speak and listen
You talk to Claude and hear its replies in a spoken voice, instead of typing and reading.
Same conversation
You can switch between text and voice within the same chat without starting over.
Connected abilities
During a voice conversation, Claude can use web search and tools you have connected, such as email or calendar.
Many languages
Voice mode supports a range of languages, with non-English support described as beta.
Typing to Claude is like exchanging letters with a pen friend. Voice mode is like picking up the telephone. The knowledge on the other end is the same, but the experience is quicker, more natural and easier when your hands are busy.
Voice mode is not a separate assistant. It is another way to talk to the same Claude, with the conversation staying in one place whether you type or speak.
2Voice Mode Versus Dictation
Two features sound alike but do different jobs, and mixing them up is a very common mistake.
Dictation turns your speech into typed text. You talk, the words appear in the message box, and then you send the message just as if you had typed it. Claude replies in text. Dictation is simply a faster way of typing.
Voice mode is a full two-way spoken conversation. You speak to Claude, and Claude speaks back. The reply arrives as audio, not only as text on a screen.
| Feature | You do | Claude does | Best for |
|---|---|---|---|
| Typing | Type your message | Replies in text | Precise wording, code, long documents |
| Dictation | Speak, and your words become text | Replies in text | Quickly entering a long prompt |
| Voice mode | Speak | Replies in spoken voice | Natural conversation, busy hands |
Dictation is like dictating a letter to a secretary who types it up for you. Voice mode is like having a phone call. In the first case you still read the answer. In the second, the answer comes back in a voice.
In some Claude work environments built for longer technical tasks, dictation is available but voice mode is not. Check the official help page for where each feature works.
3How a Spoken Conversation Works
There is a surprising amount of work hidden inside a single spoken question and answer.
When you speak, your voice is captured by the microphone as a sound signal. A part of the system converts that sound into written words, a step often called speech recognition. The text is then passed to Claude, the language model, which works out a reply in words. Finally, a different part converts those written words into sound, a step called speech synthesis or text-to-speech. You then hear the result through the speaker.
The exact engineering can vary and is not something users need to manage, but this simplified picture is enough to understand why certain things happen. If the words are not recognised correctly in the first step, Claude may answer a slightly different question from the one you asked. If you are in a noisy place, the first step has more trouble. And the quality of the spoken voice depends on the final step.
flowchart LR
A["You speak"] --> B["Microphone captures sound"]
B --> C["Speech becomes text"]
C --> D["Claude reads and thinks"]
D --> E["Reply written as text"]
E --> F["Text becomes a spoken voice"]
F --> G["You hear the answer"]
Imagine a conversation through two interpreters. One listens to you and writes down what you said. The expert reads the note and writes a reply. A second interpreter then reads the reply aloud to you. Each person in the chain matters, and a slip at any stage affects the final result.
Knowing this chain explains most practical advice about voice mode: speak clearly, reduce background noise, and check that the words Claude seems to have heard match what you meant.
4Two Ways to Talk: Hands-Free and Push-to-Talk
Voice mode offers different styles of turn-taking, so you can match it to your surroundings.
In hands-free mode, Claude listens continuously and decides when you have finished speaking. You do not need to press anything between turns. This feels the most like a real phone call and works well when you are cooking, walking or driving, as long as the environment is reasonably quiet.
In push-to-talk mode, you control when Claude listens. You press or hold a control while you speak, and release when you are done. This is useful in noisy places, or when you want to pause to think without Claude assuming you have finished. Think of it like a walkie-talkie, where you press the button to speak and release it to listen.
Advantages
- Hands-free feels natural and keeps your hands free for other tasks
- Push-to-talk gives you full control over when Claude listens
- Push-to-talk reduces accidental pickup of background chatter
- You can choose whichever style suits the moment
Disadvantages / Trade-offs
- Hands-free can be less reliable in noisy places
- Hands-free may cut in if you pause while thinking
- Push-to-talk needs a hand or finger on the screen
- Which options appear can depend on your device and environment
Hands-free is like chatting with a friend across a quiet table, where both of you sense when it is the other’s turn. Push-to-talk is like a formal meeting where you raise a hand and wait to be called on. Each has its place.
If Claude keeps jumping in before you finish a thought, try a quieter spot, speak in steadier sentences, or switch to push-to-talk so you decide when your turn ends.
5Getting Started in Simple Steps
The exact buttons change as the apps are updated, so this is a general picture of the process.
Open Claude
Start a chat in the Claude mobile app, Claude Desktop or the web. Voice mode is documented as working best from a phone.
Start voice mode
Look for the voice control in the chat area and tap or click it to begin a spoken conversation.
Allow microphone access
Your device may ask permission to use the microphone. Voice mode cannot hear you without it.
Choose a voice and pace
Claude offers several preset voices, and settings let you adjust the pace, so you can pick what sounds comfortable.
Speak naturally
Ask your question in everyday language. If Claude starts going in the wrong direction, you can simply speak up and it will stop and listen.
Leave voice mode when finished
Exit when you are done. The conversation can continue in text if you wish.
Since menus and button positions are updated from time to time, the official “Use voice mode” article on support.claude.com is the most reliable guide to the exact steps on your device.
Starting voice mode is like walking into a room and picking up a phone. You make sure the phone works, you choose a comfortable volume, and then you simply start talking.
6Which Claude Models Power Voice Mode?
Speed and depth pull in opposite directions, and voice mode has had to balance them.
In its early days, voice mode ran only on Claude Haiku, the model chosen for speed. This kept conversations quick, which is very important for spoken exchanges, because long pauses feel awkward in conversation. The trade-off was that quick did not always mean deep.
Anthropic later announced that voice mode runs on Claude Opus, Claude Sonnet and Claude Haiku. That means hard questions can now be thought through with the more capable models, not only quick ones. The models you can use in voice mode follow your plan, and a model selector inside voice mode lets you switch models during a conversation. Voice conversations are described as starting on whichever model you last selected in text chat.
| Model tier | Strength in voice | Good for |
|---|---|---|
| Haiku | Very quick replies | Short questions, casual chat, quick lookups |
| Sonnet | Balance of speed and quality | Everyday explanations, planning, general advice |
| Opus | Deeper thinking on hard problems | Complex reasoning, detailed discussions |
At the time of writing, the top-tier Claude Fable model is documented as not yet available in voice mode. When picking a model in voice mode, you choose a tier without choosing a specific generation. Check the help page for the current list.
Imagine three advisers on call. One answers in a heartbeat but keeps things brief. One gives balanced answers at a steady pace. One takes a moment longer but works through tough puzzles carefully. You can ask for whichever suits the question.
flowchart TD
A["Your spoken question"] --> B{"How hard is it?"}
B -- "Quick and simple" --> C["Haiku"]
B -- "Everyday depth" --> D["Sonnet"]
B -- "Hard problem" --> E["Opus"]
C --> F["Spoken answer"]
D --> F
E --> F
7Tools and Search During a Voice Chat
Voice mode is not limited to what Claude already knows. It can reach out for fresh information and your own data.
In voice mode, Claude can use web search to look up current information, so it is not limited to what it learned during training. It can also use tools you have connected, such as Gmail, Google Calendar, Google Docs and Slack. This allows spoken requests like asking about your schedule or having an email thread summarised out loud.
A connected tool is an outside service that you have given Claude permission to access. Without your connection, Claude cannot see your email or calendar. The connection is something you set up and control, and what Claude can reach depends on what you have allowed.
Checking your day
While getting ready in the morning, you might ask Claude out loud what is on your calendar, and hear a spoken summary without touching a screen.
Catching up on messages
Walking to the station, you could ask for a spoken summary of an email thread so you arrive already briefed.
Looking up fresh facts
When you ask about something recent, Claude can search the web and speak the result, rather than relying only on older knowledge.
Thinking out loud
You can talk through a plan or a decision, and Claude can respond, ask questions and help you organise your thoughts.
Spoken answers can be heard by anyone nearby. If you ask Claude to read out private emails or personal details, think about who is around, or use headphones.
Think of a personal assistant who not only knows a lot but can also look things up in the library and flip through the files you have handed over. The more you allow, the more they can help, and you decide what they can open.
8Languages and Voices
Speaking is personal, so choices about language and voice matter more than they do in text.
Voice mode began in English and has since been extended to many more languages. Support for languages other than English is described as being in beta. One report on the expansion listed languages such as French, German, Hindi, Indonesian, Italian, Japanese, Korean, Portuguese and Spanish, and Anthropic has since described further improvements to how natural non-English speech sounds. The list may have grown since then, so confirm the current set on the official help page.
There is one detail that surprises many people. Changing the display language of the app does not by itself change your voice language. The language you speak in voice mode is set separately in the voice settings. You can also ask Claude to switch languages in the middle of a voice conversation, which is handy if you are practising a language or speaking with someone who prefers another one.
Voice language
Set the language you will speak and hear. It is separate from the app’s display language.
Voice style
Pick from several preset voices to find one that sounds pleasant to you.
Pace
Adjust how fast Claude speaks, slower for careful listening or faster to save time.
Switching on the fly
Ask Claude to change language during a conversation when you need to.
Think of the menu language in a restaurant and the language the waiter speaks to you. They are two separate things. You can read the menu in one language and still chat with the waiter in another. Voice settings work the same way.
Claude’s voices are a limited set of preset voices and are designed to avoid mimicking specific individuals, and the standard usage policy applies in voice mode just as it does in text.
9Claude in the Car with Apple CarPlay
One of the most practical places to talk instead of type is behind the wheel.
On iPhone, voice mode now works with Apple CarPlay, the system that shows approved phone apps on a compatible car’s display. Apple opened CarPlay to voice-based conversational apps with a recent iPhone software update, and Claude joined that group. According to Anthropic’s announcement and press coverage, the idea is that you can ask Claude something while driving and keep your hands on the wheel.
The CarPlay experience is described as voice-first and fairly simple. You open Claude from the CarPlay screen, then start a new voice conversation or resume a recent chat. Reports indicate that Claude does not have a wake word like the phrases used by built-in assistants, so you need to open the app first. They also say that Claude on CarPlay does not control the car or other functions of your iPhone. It is there to talk with, not to operate the vehicle.
Connect your iPhone to the car
Use your car’s CarPlay connection, wired or wireless, depending on your vehicle.
Open Claude from the CarPlay screen
Select the Claude icon among your CarPlay apps. A first tap is needed, as there is no spoken wake word.
Start or resume a conversation
Begin a new voice chat, or pick up a recent one where you left off.
Talk and listen
Ask your question aloud and listen to the reply while your eyes stay on the road.
Driving needs your full attention. Keep conversations short, avoid complicated tasks while the car is moving, and follow your local laws about device use. If a topic needs deep concentration, wait until you have stopped.
Claude in the car is like a well-informed passenger who is happy to chat, answer questions and help you think. A passenger cannot steer, change the radio on their own or press buttons for you, and neither can Claude here. They simply keep you company with their knowledge.
10Everyday Uses for Voice Mode
Voice shines whenever your hands, eyes or patience for typing are limited.
Hands busy at home
While cooking, you can ask for substitutions, timings or a step you forgot, without touching the screen with messy fingers.
Commuting and walking
Use travel time to talk through an idea, rehearse a presentation, or learn about a topic that interests you.
Language practice
Speaking a new language aloud with Claude gives you a patient practice partner, and you can ask it to slow down or explain a phrase.
Brainstorming and planning
Talking is often faster than typing for rough ideas. You can think aloud and let Claude organise and challenge your thoughts.
Accessibility
For people who find typing difficult or reading from a screen tiring, speaking and listening can make Claude far easier to use.
Interview and speech practice
You can rehearse answers aloud and ask Claude for feedback on clarity and structure.
flowchart TD
A["Voice mode"] --> B["Hands busy"]
A --> C["On the move"]
A --> D["Learning and practice"]
A --> E["Thinking aloud"]
A --> F["Accessibility"]
A kitchen timer, a notebook and a recipe book each have their place. Voice mode is like having a helpful companion in the room when your hands are full. You use it when speaking is simply the easiest option.
11Tips for Better Voice Conversations
A few small habits make a big difference to the quality of what you get back.
Choose a quiet place
Less background noise means more accurate recognition. When quiet is impossible, push-to-talk helps.
Speak in complete thoughts
Clear, full sentences give Claude more to work with than fragments.
Give context up front
Say what you are trying to achieve, for example that you want a simple explanation or a short answer.
Ask for the format you want
Because answers are spoken, you can ask for short replies, step-by-step guidance, or a summary of the main points.
Interrupt when needed
If an answer drifts off course, speak up. Claude will stop and listen, so you do not have to sit through it.
Switch to text for precision
For long lists, exact numbers, links or code, move to the text view of the same conversation, where details are easier to read and copy.
Voice is excellent for ideas and conversation, but text is better for things you need to see exactly, such as an address, a figure or a list of steps you will follow while doing something else.
Good telephone manners apply here. Speak clearly, say what you want, and do not be shy about asking the other side to repeat or slow down. The better you communicate, the better the conversation goes.
12Limits, Usage and Access
Voice mode is useful, but it has boundaries that are worth knowing before you rely on it.
Usage limits
Voice conversations count toward the regular usage limits of your subscription plan, just like text chats.
Beta status
As a beta feature, voice mode may change, and some behaviour may be less polished than text.
Model availability
Models available in voice follow your plan, and some, such as Fable at the time of writing, are not yet supported.
Where it works
Voice mode works in the main Claude apps but is not available in every Claude work environment, such as ones aimed at long technical tasks.
| Topic | What the documentation indicates |
|---|---|
| Plans | Available on Free, Pro, Max, Team and Enterprise plans |
| Devices | iPhone, Android, Claude Desktop and the web, built to work best on phones |
| Usage | Counts toward your normal plan limits |
| Organisations | Enterprise owners can ask Support to disable voice mode for their organisation |
| Car use | iPhone with Apple CarPlay, voice-first, no control of the car or phone |
Voice mode can still make mistakes, just like text. A confident spoken answer is not automatically a correct one. For health, money, law or safety, verify important points with trusted sources or professionals.
Like a mobile phone plan with a monthly allowance, a long chat uses up some of your share, whether you type or talk. If you hit the limit, it is not a fault; it is simply the allowance running out.
13Fixing Common Problems
When something does not work, a short checklist usually solves it.
Problem
Claude stops responding in the middle of a voice conversation.
Why It Happens
Common causes include reaching your usage limit, an unstable internet connection, or a temporary glitch in the session.
Correct Approach
Check whether you have reached your usage limit, make sure your connection is steady, then leave voice mode and enter it again.
Problem
Claude does not seem to understand what you said.
Why It Happens
Background noise, a distant microphone, or very quiet speech can make the speech recognition step less accurate.
Correct Approach
Move somewhere quieter, bring the phone closer, speak a little more clearly, or use headphones with a microphone. Rephrase if the answer suggests Claude misheard.
Problem
Claude keeps interrupting you before you finish.
Why It Happens
In hands-free mode, a pause for thought can look like the end of your turn.
Correct Approach
Use push-to-talk so you control your turn, or plan your sentence before speaking.
Problem
Voice sounds like the wrong language or accent.
Why It Happens
The voice language is set separately from the app’s display language, so changing one does not change the other.
Correct Approach
Open the voice settings and choose the language you want, or ask Claude to switch languages during the conversation.
If a problem persists, the Claude help centre at support.claude.com has an article on voice mode with current troubleshooting advice, and the feedback buttons in the app let you report issues.
14Benefits and Trade-offs of Talking to Claude
Voice is a wonderful addition, but it is not always the best tool.
Advantages
- Natural, conversational feel that is easy to learn
- Works when hands or eyes are busy
- Often faster than typing for ideas and questions
- Helpful for accessibility and for language practice
- Seamless switching between speaking and typing
- Can use search and connected tools while you talk
Disadvantages / Trade-offs
- Noise and accents can reduce recognition accuracy
- Not private in shared spaces
- Harder to scan, copy or reread than text
- Spoken answers favour shorter, simpler responses
- Still a beta feature with some limits
A phone call is perfect for a quick chat, but you would not use it to send a contract. Voice mode and text each have jobs they do best, and the skill is knowing which one to reach for.
15Frequently Asked Questions
Yes. It is documented as a beta feature available on all plans, including Free, Pro, Max, Team and Enterprise. Voice conversations still count toward your plan’s usage limits.
On the Claude mobile apps for iPhone and Android, on Claude Desktop and on the web, though it is described as built to work best from a phone. On iPhone it also works with Apple CarPlay.
Dictation converts your speech into typed text, and Claude replies in text. Voice mode is a full two-way spoken conversation, where Claude speaks its replies aloud.
Voice mode runs on Claude Opus, Sonnet and Haiku, and the models you can use follow your plan. You can switch models during a conversation, though the top-tier Fable model was not yet supported at the time of writing.
Yes, if you have connected those tools, such as Gmail, Google Calendar, Google Docs or Slack. Claude can also use web search. Without your connection, it cannot access your accounts.
Yes. Voice mode supports many languages, with non-English support described as beta. Set your voice language in the voice settings, and you can also ask Claude to switch languages mid-conversation.
Yes. If Claude is heading in the wrong direction, you can start talking and it will stop and listen.
No. Reports describe it as a voice-first conversation tool. You open it from the CarPlay screen, and it does not control vehicle functions or other iPhone features.
Dictation is available in those environments, but voice mode is not. Check the official help page, because availability can change.
Enterprise owners who want to disable voice mode for their organisation can contact Support to arrange it.
The official Claude help centre at support.claude.com has an up-to-date article on using voice mode. Because features and menus change, it is the best source for exact steps and the latest availability.
16Summary and Key Takeaways
Voice mode lets you have a complete spoken conversation with Claude. Your speech is turned into text, Claude works out a reply, and that reply is spoken back to you. You can choose between hands-free and push-to-talk, pick a voice and pace, interrupt when needed, and move between speaking and typing in the same chat. It runs on Claude Opus, Sonnet and Haiku, follows your plan’s model choices, can use web search and connected tools, supports many languages in beta, and now reaches into the car through Apple CarPlay on iPhone. Like any beta feature, it has limits: usage counts against your plan, noise can affect accuracy, and answers can be wrong, so important facts deserve a second check. Used for what it does best, which is natural conversation, quick questions and hands-busy moments, it is a very convenient way to work with Claude.
Key Takeaways
- Voice mode is two-way speech — you speak and Claude speaks back, unlike dictation, which only turns your words into typed text.
- It is a beta feature on all plans — available across the mobile apps, Claude Desktop and the web, and built to work best on phones.
- Pick the style that suits the moment — hands-free for natural flow in quiet places, push-to-talk for noise or full control.
- Models and tools extend its power — Opus, Sonnet and Haiku are supported, and Claude can use web search and connected tools like Gmail, Calendar, Docs and Slack.
- Languages and voices are adjustable — voice language is set separately from the app’s display language, and you can ask Claude to switch languages mid-chat.
- CarPlay brings Claude into the car — on iPhone it is a voice-first experience opened from the CarPlay screen, and it does not control the vehicle or phone.
- Know the limits and stay careful — voice counts toward usage limits, background noise affects accuracy, and spoken answers still need checking for anything important.