Transcribe WhatsApp voice messages
Export a chat from WhatsApp and drop the ZIP file in. Dikdukai finds every voice message, transcribes it and puts them all into one document, with the sender and time above each message.
New in 1.0.4 · Offline document translation
Drop in a recording, a video or a WhatsApp chat and get a clean transcript with speaker identification, ready for Word and PDF. It works in Hebrew, English, Arabic and Russian, and can also translate the document. 100% offline processing: transcription and translation run on your computer, and your recordings never leave it.
What it does
Dikdukai is professional transcription software for Windows, built first for Hebrew. It converts audio to text, organizes the conversation by speaker and produces a document you can send, in another language if you need it.
Export a chat from WhatsApp and drop the ZIP file in. Dikdukai finds every voice message, transcribes it and puts them all into one document, with the sender and time above each message.
Speaking Hebrew but need the document in English? Choose a document language and the transcript is translated into Hebrew, English, Arabic or Russian, with the speakers and timestamps. Translation runs on your PC, and the original is kept.
Speakers are separated automatically. Click "Speaker 1", type "Attorney Cohen", and the name updates across the whole transcript at once.
Every line has a timestamp. Click a line's timestamp to play that part of the recording, so checking and fixing is easy.
Amounts, dates, times, phone numbers and addresses mentioned in the conversation are collected in one panel. Click an item to jump to the line where it was said.
Export a tidy Word or PDF document with your office logo and name at the top, or an Excel sheet. You can also export plain text (TXT) and SRT subtitles for videos.
Record a meeting, a lesson or a dictation right in the app, then click "Transcribe Now". The recording is saved on your PC only.
Each transcript is saved in its client or project folder. You can also set a watched folder: every recording saved there is queued automatically.
Search for a word or a name and find it in every transcript you have made, in one organized library.
Transcription and translation run on your own computer. The recording, the transcript and the document are never sent to any server. Suitable for sensitive material.
The installer is about 220MB. During setup you choose which components to download, and you can add more later by running it again.
When a new version is out, a message appears. One click, and the app updates itself and reopens. Your transcripts, settings and subscription are kept.
New in version 1.0.4
Speaking one language and need the document in another? In Dikdukai you choose the spoken language and the document language separately, and the transcript is translated on your own computer, with no internet and no cloud.
The language spoken in the recording: Hebrew, English, Arabic, Russian, or "Mixed" for a conversation that switches between languages. It picks the transcription engine.
The language your files are written in: Hebrew, English, Arabic or Russian, even when the recording is in another language. In a mixed conversation, parts in other languages are translated into the document language.
The speakers, timestamps and line structure carry over to the translation. Headings and tables in the document are also in the chosen language, with the correct text direction (right to left for Hebrew and Arabic).
In the editor, choose what to show: the original, the translation, or both. Double-click a line to correct the translation, and the "Translate" button translates older transcripts too.
You can get documents in the recording's language as well as translated ones: Word, PDF, Excel, TXT and SRT subtitles.
Translation runs on your PC, just like transcription. The recording, the transcript and the translation are never sent to any server.
Good to know: translated documents are marked "Machine translation" and are not a certified translation. They are meant for understanding, so check them against the original, especially names, numbers and times. Lines worth checking are marked in the editor and in the document. Translation adds processing time: roughly 40 to 50 minutes per hour of recording on a powerful computer, and longer on a slower one. It requires a computer with at least 8GB of memory and the "Offline document translation" component at installation.
Installation, updates and privacy
Setup downloads only what you choose, and new versions install from inside the app, in one click.
The installer is about 220MB. During setup you pick the components, and it downloads only those, with a progress bar. If the connection drops, what was already downloaded is kept and the download resumes from the same point. You can add more components later by running the installer again.
When a new version is out, a message appears on the home screen. Click "Update now" and the app downloads the update, installs it and reopens. Your transcripts, settings and subscription are kept. Windows may ask for your approval.
Your recordings and transcripts never leave your computer. The app only checks once a day whether a new version is available, without any information about you or your files (as with any website visit, the site sees your IP address). You can turn this off in Settings, under "Check for updates automatically".
See it
A clean interface in English, Hebrew, Arabic and Russian, with light and dark modes. The screenshots below show the Hebrew interface with sample data.




How it works
A recording, a video, a whole folder or a WhatsApp chat. You can also record straight from your microphone.
Choose the spoken language (or "Mixed" for a conversation in several languages) and, if you need a translation, a different document language. Everything runs on your computer, without internet.
Name the speakers, fix anything you need and export to Word, PDF, Excel or SRT.
Who it is for
A podcast, an interview, a private lesson, a lecture, a meeting or a client call: every video and every recording becomes text, with timestamps. (The video is in Hebrew.)
Transcribe hearings, mediations, meetings and client calls. Sensitive material stays in the office.
Transcribe recorded lectures and lessons into searchable notes.
Turn a recorded lesson into text and subtitles for the video.
Transcribe interviews and calls, with marked quotes and timestamps.
Transcribe episodes and videos, and export SRT subtitles for YouTube and social media.
Transcribe job interviews, disciplinary hearings and committee meetings for the minutes.
Amounts, dates and phone numbers from calls, all in one place.
14 voice messages from a client? One document, with a name and time for each.
Download and install
The installer is small, about 220MB. During setup you choose which components to download, so installation needs an internet connection. After that, transcription and translation run on your PC with no internet.
Components you can choose during setup:
Recommended: full Hebrew transcription (Hebrew fast and accurate modes, plus speaker detection) is about 4.7GB, and Everything is about 12GB. To transcribe English, Arabic or Russian, or mixed conversations, choose "Everything" or add "More languages" in a custom selection.
If Windows blocks the file, there is another way: before running it, right-click it, choose Properties, tick Unblock and click OK. To add a component later, run the installer again and select it: components that are already installed are not downloaded again. For an office with no internet access at all, a full offline version is available on request by email.
Requirements: Windows 10 or 11 (64-bit) · internet during installation · free disk space during installation: about 10GB (Recommended) or about 17GB (Everything); once installed, the components take about 4.7GB or 12GB · 16GB RAM recommended (8GB works in Fast mode; translation needs at least 8GB)
Free pilot
After installing you can try it right away, no sign-up: one file up to 3 minutes. To get 30 days free with no limits, email us your installation code and we will send you an activation key.
In the app, click the gear ⚙ at the bottom of the side menu.
In the "Installation code" row, click "Copy code". It is a long code of letters and numbers, and it is not personal information: it only links the activation key to your computer.
Paste the code into an email to dikdukai@gmail.com. You can use the ready-made template below.
When the key arrives, click "Enter activation key" in Settings, paste it, and your pilot starts.
Send your installation code to:
Email template (Subject: Request to join the Dikdukai pilot)
Hello, I would like to join the free 30-day Dikdukai pilot. My installation code: (paste here) Full name: Occupation: Phone (optional): Thank you!
Transcription converts speech to text: a recording, a call or a video becomes written text. Most automatic transcription services work in the cloud, which means your recording is uploaded to someone else's server. Dikdukai does the same job locally on your Windows PC: no internet, no upload, so the recording, the transcript and the document stay with you.
Instead of listening to dozens of voice notes one by one and typing them out, use WhatsApp's "Export chat" and drop the ZIP file into Dikdukai. It finds the voice messages, transcribes them and puts them in order, with the sender and the time. Works with chats exported from Android and iPhone. If the messages are in another language, you can get the document translated.
In a conversation with several people, Dikdukai separates the speakers and labels them "Speaker 1", "Speaker 2" and so on. Name each speaker once and the name updates throughout the transcript. The result reads like proper minutes of a meeting, hearing, mediation or interview.
Dikdukai transcribes video files (MP4, MOV, MKV and more), not just audio. Students transcribe lectures, teachers turn recorded lessons into text, and creators export SRT subtitles to upload with their videos to YouTube. You can also get translated subtitles in another language, and the first subtitle notes that they are a machine translation.
Choose the spoken language, the language used in the recording, or "Mixed / auto-detect" for recordings that switch between languages. Separately, choose the document language: the document can stay in the recording's language or be translated into Hebrew, English, Arabic or Russian. Hebrew transcription is where Dikdukai is strongest, using models trained on Hebrew, with an extra Accurate mode for Hebrew. The interface is available in all four languages.
Have a Hebrew meeting and need the minutes in English? Or an Arabic or Russian recording you need to understand? Choose a document language in Dikdukai and the transcript is translated on the computer itself, without sending it to a cloud translation service. The speakers and timestamps are kept, you can get both the original and the translation, and even translated SRT subtitles. Translated documents are marked "Machine translation": they are meant for understanding and are not a certified translation.
A recorded phone call, a Zoom meeting recording or any audio file: drop it into Dikdukai and get a transcript. You can also drop a whole folder of recordings, and they are queued and transcribed one after another.
Law firms and HR teams work with sensitive recordings: disciplinary hearings, mediations, committee meetings and client meetings. With Dikdukai the recordings never leave the office, each speaker gets a name, and the document carries your firm's logo. Keep in mind: this is automatic transcription, not a certified court transcript.
If you are looking for a Windows alternative to cloud tools like Otter or TurboScribe, or a desktop app in the spirit of MacWhisper, Dikdukai runs entirely on your PC, translation included. That matters for lawyers, HR teams, researchers and anyone handling confidential recordings.
Questions and answers
Download, try it right away, and join the free 30-day pilot.