f4 Online Manual
As soon as transcription mode is active, you can automatically convert audio and video files to text in the bottom section of the left-hand column.
f4x speech recognition is an online service, i.e. an Internet connection is required to use it. An f4x account is also required for use and, if necessary, an AVV (order data processing). This can be created free of charge. Click on the “Login” button to log in or register.
You need sufficient quota credit for automatic transcription. After your first registration you will receive a free trial quota. The available quota is displayed at the bottom left. The number of quota minutes corresponds to the amount of data material that you can upload and automatically convert into text.
The quota can be increased by clicking on the three dots next to the quota display and redeeming your purchased quota code here. If you no longer have a valid code, click on the link to our online store to buy more credit.
For automatic recognition, click on the plus symbol in the left-hand column and select “Automatic transcript”. You can then select the audio and video file(s) from your computer.
After selecting, you can set the language of the recording and determine how time marks and pauses are to be marked. In a word list, you record specific technical terms that occur in the media files (all separated by commas). This increases the recognition accuracy for these terms.
Note: A file in which different languages are spoken will not be recognized correctly.
Uploading and editing takes a few minutes. German-language media are about a tenth of the media length, all other languages require about half the media length for processing.
Notes:
Once the speech recognition on the server is complete, a button appears next to the corresponding text “Accept text”. When you click on this button, the text file is downloaded, transferred to the project and deleted from the server. It is now permanently integrated in your project file.
Please note: Errors are quite likely to occur with automatic transcription. For example, A word is not recognized or is recognized incorrectly. Similarly, the speaker’s name may be incorrect or missing. Too many speakers or none at all (especially when speakers are similar or their roles overlap). You can correct all of this manually in f4 after it has been detected. Uploading the file again will NOT produce a different result and will only use up your quota minutes unnecessarily.
The transcripts are uploaded to our servers exclusively for speech recognition. The transmission is encrypted. Opening the transcripts from the server is only possible from the computer from which the files were uploaded.
Once the files have been successfully imported into the project, the data will only be available locally on your computer. Your data will NOT be stored on our server.
If you wish to send us personal data belonging to third parties (such as people interviewed), the GDPR requires that a so-called data processing agreement be entered into beforehand.
In this contract, we only guarantee appropriate security measures and things such as rights to issue instructions and deletion periods are agreed. We have prepared a corresponding contract. You can complete this process under the “Manage Account” option, which will redirect you to a webpage where you can log in with your account information and finalize the contract.
The AI functions (AI Impulse and AI Dialog; see Chapter 5) also send the selected text to a server for processing. The requests are processed via the servers of GWDG (Gesellschaft für wissenschaftliche Datenverarbeitung mbH Göttingen), a scientific institution based in Germany.
You can find the complete terms and conditions in our Terms of Service for Speech Recognition and AI Processing at https://www.audiotranskription.de/spracherkennungsvertrag/.
The contract for commissioned data processing described in section 2.4.2 also covers the use of AI functions.
