You can ask a question aloud on your iPhone, hear an AI answer, and continue speaking without using the microphone button for each turn. OGAM (Off Grid AI Mobile) Pro provides this through its Audio interface and Hands-free turn setting. Download the local models and voice assets first. The conversation can then run without internet.
Get OGAM on the App Store | Mobile features and requirements
This guide sets up an active conversation inside OGAM. Keep the app visible for the first test. It does not configure Siri, a system wake word, or a voice assistant for other iPhone apps.
You can use it to talk through a plan, ask for a short explanation, or continue a discussion while your phone rests on the desk.
What does your iPhone need?
The app's published requirements include iPhone 12 or newer on iOS 17 or later. For the full spoken conversation, you need OGAM Pro and local models for recognition, answers, and voice output. You also need microphone permission and enough free storage for those downloads.
Prepare all of these while connected:
| Component | What it provides |
|---|---|
| Local transcription model | Converts your spoken question into text |
| Local chat model | Generates the answer |
| Local voice model and selected voice assets | Speaks the answer |
| Pro setup | Provides the Audio interface and speech-output feature |
Start with smaller models. The iPhone must make room for the chat and speech workloads. A large chat model that runs by itself may leave too little memory for a useful voice conversation.
Check the installed App Store version. Public GitHub releases can appear before an App Store update. Use the current iPhone app and check for updates if the hands-free controls are missing.
How do you prepare the three local models?
Download an on-device transcription model in Model Settings > Transcription (Speech to Text) > Transcription model. Select a downloaded local chat model for answers. Then download the local voice model in Models > Voice and wait for it to become ready.
For a first multilingual setup, Base, 99 languages is a small speech-recognition option. English-only variants do not become multilingual when you change the phone's language.
Set Language in the speech-to-text settings to the language you plan to speak. In the text-to-speech settings, select a voice for the language you want to hear.
Speech input and output do not share one model. Selecting a French transcription language does not by itself download a French speaking voice.
If a voice selection starts another download, wait for the ready message. A working English voice does not establish that all other language assets are stored on your iPhone.
Where do you enable Hands-free on iPhone?
In Model Settings > Text to Speech, set Interface Mode to Audio. Then open Transcription (Speech to Text) and set Voice turns to Hands-free. The same voice-turn controls are available in Chat Settings under SPEECH TO TEXT.
Choose the mode that matches what you want:
| Voice turns | You control |
|---|---|
| Manual | When recording starts and stops |
| Auto | When recording starts; silence ends it |
| Hands-free | The conversation proceeds through detected speech and pauses |
Hands-free is for the Audio interface. It does not turn ordinary text-chat dictation into automatic listening.
For a first conversation, use End of turn: 5s and Listen again: 2s. These are the current defaults.
End of turn controls how long a pause ends your question. Listen again controls the wait after a reply before the microphone reopens. Start with enough time to finish a thought, then adjust the delay to suit your speaking pace.
How do you start an offline conversation?
Finish model setup, allow microphone access, and enter Audio mode with Hands-free selected. When the app is waiting for your voice, speak a short question and pause. The app transcribes it, sends the recognized turn to the local chat model, and plays the answer using the prepared local voice.
1. Prepare the iPhone's audio
Put the iPhone close enough to hear you. Use a quiet room for the first check and set a comfortable media volume.
Allow microphone access when iOS asks. If permission was denied earlier, enable it for OGAM in iPhone settings before trying again.
Check the audio output if you have Bluetooth headphones connected. You may hear the reply there rather than from the phone's speaker.
2. Turn off the connection
After downloads and Pro setup are complete, enable airplane mode. Check that Wi-Fi is off as well.
Use an ordinary conversation for this check. A request for fresh web information introduces a separate network requirement.
3. Ask a short question
In an empty Audio conversation, Waiting for your voice means the app is listening. If it instead shows Tap to speak, use the microphone control to start.
Try:
Help me plan a short presentation. Ask me one question at a time.
Pause for the selected end-of-turn period. The recognized question is submitted automatically in Audio mode. There is no separate text-review step before each turn is sent.
4. Answer the follow-up
Listen to the reply, then wait for listening to resume. Give a short answer and let the next turn finish.
This tests the full loop. A single successful recording only checks part of the setup; the next turn also checks that recording and playback hand control back correctly.
5. End the voice session
Change Interface Mode back to Chat when you are finished. Leaving Audio mode cancels an active recording and resets the voice session.
Watch the app's recording state. The empty conversation distinguishes Waiting for your voice from Recording you now. Use those states to tell whether the app is listening or capturing a spoken turn.
How do you stop a reply and speak again?
Use the square Stop control while the assistant is generating or playing a reply. Then use the microphone control when it is available to begin your next turn. A deliberate stop can also stop automatic listening, so wait for a visible listening or recording state before speaking.
Do not rely on speaking over the reply. In release 0.0.111, recording waits while the assistant speaks to prevent its own voice from becoming the next input. Earlier release notes described talk-over interruption, but that is not the behavior to expect from this release.
How do you make the pauses fit your speaking pace?
If the app submits your question while you are thinking, increase End of turn. The available choices are 1s, 2s, 3s, and 5s. If you need a longer uncontrolled pause, use Manual and stop the recording yourself.
If the phone picks up the end of its own reply, reduce speaker volume or increase Listen again. Its choices are 0.5s, 1s, 2s, and 3s.
A shorter delay is not automatically better. Choose settings that let you finish the sentence and prevent the reply from becoming the next input.
What should you check when it does not work?
| Problem | Check | Action |
|---|---|---|
| Hands-free does not start automatically | Interface mode | Select Audio as well as Hands-free. |
| Microphone input never begins | iOS microphone permission | Allow microphone access for the app, then retry. |
| Speech is recognized but no answer arrives offline | Chat model selection | Select a downloaded local chat model. |
| A text answer appears but no voice plays | Voice readiness and audio output | Finish downloading the selected voice and check media volume and output routing. |
| The question ends during a pause | End of turn | Choose 5s or switch to Manual. |
| The answer is unrelated | Recognized question | Check the transcript, speech language, and microphone position. |
| Switching languages breaks offline playback | Missing assets for the new voice | Reconnect, prepare that language, then repeat the offline test. |
| A model cannot load | Free memory | Use smaller models and close other memory-heavy apps. |
Background speech and music can keep the audio level above the silence threshold. If the environment is noisy, use Manual control and check the transcript rather than relying on automatic pause detection.
Does the conversation leave the iPhone?
When transcription, chat, and voice output all use downloaded local models, those processing steps run on the phone. App downloads, Pro activation, and model setup are separate connection-dependent steps.
A remote speech service, remote chat model, or connected tool changes that setup. Keep the first test local, with a simple prompt and airplane mode enabled.
iPhone hands-free FAQ
Do I need Pro?
Yes, for the full Audio interface and spoken replies described here. Basic speech-to-text is a separate capability.
Is the phone always listening?
This guide covers the app's active voice session and visible listening state. It does not promise a permanent background listener, lock-screen operation, or a system wake word.
Can I choose a different language for answers?
You can ask the chat model to answer in a supported target language and select a matching voice. Check the written reply; changing the voice language alone does not translate it.
Can I use this on a flight?
The local processing can work after setup with the required assets downloaded and Pro access available. Test the complete conversation in airplane mode before you need it away from a connection.
Try a short question and one follow-up while offline. Check the recognized words before you shorten the pauses.
Top comments (0)