Echo Node offers two ways to turn a recording into text: free on-device transcription and optional Premium transcription powered by Gemini.
Both methods create editable transcripts inside Echo Node. The main differences involve where processing happens, whether an internet connection is required, the available transcription capabilities, and whether Premium usage is consumed.
Neither option is perfect or universally better for every recording.
On-device transcription processes the selected recording locally using transcription capabilities available on your Android device.
The recording does not need to be submitted to Gemini for this process.
On-device transcription is included in Echo Node’s Free experience and can be a good choice when:
You want to keep the transcription process on your device
You do not have an internet connection
You want a quick, editable transcript
The recording is clear and relatively easy to understand
You do not need Premium cloud processing
The capabilities and performance of on-device transcription can vary between Android devices.
Gemini transcription is an optional Premium feature that sends the selected recording for cloud processing.
It is designed to produce a more capable timed transcript with enhanced speaker separation. Premium voice profiles can also help Echo Node associate recognized speech with the correct person when the recording and transcription results support it.
Gemini transcription can be a good choice when:
The recording contains multiple speakers
Speaker identification is important
The audio is more complex or difficult
You want a more polished starting transcript
You want to use Premium voice-profile capabilities
Cloud processing is acceptable for that recording
Gemini transcription requires an internet connection and uses the current Premium transcription allowance.
Processing happens locally on your Android device. The recording does not need to be submitted to Gemini to create the transcript.
The selected audio is sent for cloud processing after you deliberately request Gemini transcription.
Echo Node does not automatically submit every recording to Gemini.
On-device transcription can work without sending the recording to an online transcription service.
Availability and performance depend on the transcription capabilities supported and installed on your Android device.
Gemini transcription requires an internet connection because the selected audio must be submitted for cloud processing.
Transcription accuracy varies, and neither method is guaranteed to reproduce every word correctly.
Gemini transcription is designed to provide a more capable transcription experience and may produce better results for complex recordings, multiple speakers, or challenging audio.
On-device transcription provides a private and accessible starting point without cloud processing, but results may contain more errors depending on the device and recording.
Accuracy for either method can be affected by:
Background noise
Microphone placement
Low speaking volume
Distance from the speaker
Overlapping speech
Accents and dialects
Specialized terminology
Names and uncommon words
Echoes or poor room acoustics
The recording quality
Important transcripts should always be reviewed.
On-device conversation transcription can separate detected speech into general speaker labels such as Speaker A and Speaker B when supported by the transcription result.
It may not reliably identify who those speakers are, and speaker separation can be affected by similar voices, overlapping speech, short responses, background audio, and recording quality.
Gemini transcription can produce timed, speaker-separated results and is designed to offer more capable speaker handling.
Premium voice profiles can help Echo Node associate recognized speech with enrolled people. Voice profiles improve the information available for speaker labeling, but they do not guarantee perfect identification.
Users should review speaker labels before relying on them.
Yes.
Both on-device and Gemini transcripts can be edited after they are generated.
Editing allows you to correct:
Misheard words
Names
Specialized vocabulary
Punctuation
Speaker labels
Other transcription mistakes
The edited transcript remains associated with the recording in Echo Node.
Echo Node allows transcription results to remain associated with the same recording, making it possible to work with the available transcript options without changing the original audio.
Generating or editing a transcript does not alter the recording itself.
No.
Transcription creates a text result associated with the recording. It does not rewrite, shorten, or modify the original audio file.
Editing a transcript also changes only the text, not what was recorded.
Yes.
On-device transcription is part of Echo Node’s Free experience. It does not use the Premium Gemini transcription allowance.
Echo Node’s Free experience does not contain advertising.
Gemini transcription is a Premium feature.
It uses the current Premium transcription allowance. Current subscription options, capabilities, and allowances are explained on the Echo Node Free vs. Premium page and through Google Play before purchase.
Echo Node does not currently offer a Premium trial.
No.
Echo Node does not automatically submit every recording for Gemini transcription.
You decide whether to create an on-device transcript or request Gemini transcription for a selected recording.
This allows you to choose based on the content, your privacy preference, the recording’s complexity, and whether enhanced processing is worth using for that particular file.
Choose on-device transcription when:
You want local processing
You are offline
The recording is clear
You want a free transcript
A quick, editable result is sufficient
You do not want to submit the recording for cloud processing
Choose Gemini transcription when:
You want enhanced cloud processing
The recording contains several speakers
Better speaker separation is important
You want to use voice profiles
The audio is difficult or complex
You want a stronger starting point before editing
You can make this decision recording by recording. Choosing Gemini for one recording does not require you to use it for every recording.
Automated transcripts should not be treated as guaranteed records of exactly what was said.
Review and correct a transcript before relying on it for medical, legal, financial, safety, employment, academic, or other high-consequence decisions.
When exact wording matters, compare the transcript with the original audio.
On-device transcription provides a free, local option. Gemini transcription provides optional Premium processing when you want additional capability.
Echo Node lets you decide which approach fits each recording.
On-device first. Gemini when you want it.
For more information about recordings, cloud processing, and user choices, read On-Device First: How Echo Node Handles Your Recordings and the Echo Node Privacy Policy.