How good is the automatic transcript?
This depends on the recording setting and the pronunciation of the speakers. For a “perfect” setting (e.g. a scripted TV interview), we can achieve as high as 95%-97% accuracy, whereas settings such as a spontaneous meeting, rapid speakers, accents, or noisy backgrounds like a busy café may result in 40%-50% mistranscriptions or gaps.
How can I improve the transcription quality?
In general: the better the audio recording, the better the automatic transcript. A slow native speaker in a professional audio studio will be transcribed almost perfectly, whereas a meeting in a large restaurant that was recorded on an old smartphone will have lots of errors. If you can influence the recording setting, we highly recommend at least a good microphone (entry level mics suffice, starting at roughly $40) and a quiet setting with as little background noise as possible. For more details on recording devices and tips, see this blog post on improving speech-to-text results.
As a pre-emptive action, it will greatly increase the quality if you are able to remind the speakers to speak slowly and clearly.
How long does it take until my transcript is ready?
For one hour of audio, it usually takes between 60-90 minutes. You will get an e-mail once it is ready. In comparison, a manual transcriber would take 4-6 hours.
Where is my data processed?
You can choose two levels of confidentiality for your data:
– “Normal”: Your data is stored on our local infrastructure in Switzerland. The audio files are processed by external services (e.g. Google, IBM etc.), which have their servers inside and outside of Switzerland. Our contracts with them ensure that your data is deleted once it was processed.
– “Highly Confidential”: Your data never leaves our servers in Switzerland.
Note that the “Normal” mode typically gives better transcription results, since we combine the output of various external engines with our proprietary algorithms to achieve optimum results.
Can I install Interscriber on my own servers?
Yes. We can install and run Interscriber on your local infrastructure. Please get in touch with us at info@interscriber.com
What happens if several speakers talk at the same time?
Parallel speech is hard to recognize for humans, and even more for machines. At the moment, we do not have a solution that produces reliable output for parallel speech. For this reason, we are working on an extension of our editor that significantly simplifies the process to transcribe parallel speech manually.
My recording is in an audio format that you do not support; what can I do?
We support more than 25 audio and video formats, so this is very unlikely to happen. However, if you have your recording in an unsupported file format, we recommend converting it into Wave format (typical ending .wav). There are free online services for this task, e.g.
If you prefer to use a tool on your computer (e.g.if your audio contains sensitive data), you can use:
My speakers talk in a language that is not yet supported; what can I do?
We can easily add a new language to Interscriber. Just drop us an email at support@interscriber.com
Do you support Swiss German?
Not yet. But we are working on it!
We are collaborating with the Swiss Association for Natural Language Processing (SwissNLP), four Swiss universities and several Swiss companies to build a solution. We already generated the first transcriptions in “Berndeutsch”, and it worked decently well. We will now improve the quality and extend it to other dialects.
If you want to get early access to the current state of Swiss German transcription, please get in touch at support@interscriber.com
Do you offer other solutions for Speech Processing?
Yes. SpinningBytes, the company behind Interscriber, develops custom solutions for automatic speech and text processing since 2015. If you have an idea or a question, please get in touch with us at info@spinningbytes.com
I have an improvement idea; what should I do?
We are always open to new feature ideas. In fact, we very much appreciate your feedback and will do our best to incorporate your ideas as soon as possible. Just drop us an email at support@interscriber.com
Can I use Grammarly?
Yes. Just go to https://www.grammarly.com and click on the button “Add to Firefox”. After downloading it and creating an account, you can use Grammarly without any further installation.
Powered by HTML5 Responsive FAQ
How good is the automatic transcript?
This depends on the recording setting and the pronunciation of the speakers. For a “perfect” setting (e.g. a scripted TV interview), we can achieve as high as 95%-97% accuracy, whereas settings such as a spontaneous meeting, rapid speakers, accents, or noisy backgrounds like a busy café may result in 40%-50% mistranscriptions or gaps.
How can I improve the transcription quality?
In general: the better the audio recording, the better the automatic transcript. A slow native speaker in a professional audio studio will be transcribed almost perfectly, whereas a meeting in a large restaurant that was recorded on an old smartphone will have lots of errors. If you can influence the recording setting, we highly recommend at least a good microphone (entry level mics suffice, starting at roughly $40) and a quiet setting with as little background noise as possible. For more details on recording devices and tips, see this blog post on improving speech-to-text results.
As a pre-emptive action, it will greatly increase the quality if you are able to remind the speakers to speak slowly and clearly.
How long does it take until my transcript is ready?
For one hour of audio, it usually takes between 60-90 minutes. You will get an e-mail once it is ready. In comparison, a manual transcriber would take 4-6 hours.
Where is my data processed?
You can choose two levels of confidentiality for your data:
– “Normal”: Your data is stored on our local infrastructure in Switzerland. The audio files are processed by external services (e.g. Google, IBM etc.), which have their servers inside and outside of Switzerland. Our contracts with them ensure that your data is deleted once it was processed.
– “Highly Confidential”: Your data never leaves our servers in Switzerland.
Note that the “Normal” mode typically gives better transcription results, since we combine the output of various external engines with our proprietary algorithms to achieve optimum results.
Can I install Interscriber on my own servers?
Yes. We can install and run Interscriber on your local infrastructure. Please get in touch with us at info@interscriber.com
What happens if several speakers talk at the same time?
Parallel speech is hard to recognize for humans, and even more for machines. At the moment, we do not have a solution that produces reliable output for parallel speech. For this reason, we are working on an extension of our editor that significantly simplifies the process to transcribe parallel speech manually.
My recording is in an audio format that you do not support; what can I do?
We support more than 25 audio and video formats, so this is very unlikely to happen. However, if you have your recording in an unsupported file format, we recommend converting it into Wave format (typical ending .wav). There are free online services for this task, e.g.
If you prefer to use a tool on your computer (e.g.if your audio contains sensitive data), you can use:
My speakers talk in a language that is not yet supported; what can I do?
We can easily add a new language to Interscriber. Just drop us an email at support@interscriber.com
Do you support Swiss German?
Not yet. But we are working on it!
We are collaborating with the Swiss Association for Natural Language Processing (SwissNLP), four Swiss universities and several Swiss companies to build a solution. We already generated the first transcriptions in “Berndeutsch”, and it worked decently well. We will now improve the quality and extend it to other dialects.
If you want to get early access to the current state of Swiss German transcription, please get in touch at support@interscriber.com
Do you offer other solutions for Speech Processing?
Yes. SpinningBytes, the company behind Interscriber, develops custom solutions for automatic speech and text processing since 2015. If you have an idea or a question, please get in touch with us at info@spinningbytes.com
I have an improvement idea; what should I do?
We are always open to new feature ideas. In fact, we very much appreciate your feedback and will do our best to incorporate your ideas as soon as possible. Just drop us an email at support@interscriber.com
Can I use Grammarly?
Yes. Just go to https://www.grammarly.com and click on the button “Add to Firefox”. After downloading it and creating an account, you can use Grammarly without any further installation.