- When we test by reading a book or just by chatting with each other, the transcript is not as good as testing with real voicemails. Is this expected?
Nexiwave Voicemail-to-text engine is highly tuned to voicemail-like speaking behaviours and facts (for example, there should be just one speaker in the audio). It is not expected to handle non-voicemail type audio well. Your test should be done with real world voicemail audio.
- How should I test the service?
For every service release, Nexiwave performs intensive in-house tests, with tens of hundreds real voicemail messages from across North America. Nexiwave suggests you to test with at least five real world voicemail messages, which will give you a general idea of the quality of our service.
- What level of accuracy should I expect?
We all hate listening to voicemails. The goal of our service is not 100% accurate text. Our goal is to enable users to get the main idea by reading the voicemail in text, and perform next step accordingly. To achieve this goal, Nexiwave Voicemail to text produces the accurate machine transcript results for voicemail messages. You should expect over 85% accuracy overall. Callback numbers are unarguably the most important information in voicemails. Our engine recognizes phone numbers at over 99% accuracy. Backed by these two accuracy numbers, our survey has shown that our service has consistently enabled our users to reach the goal of our service.
- Most of our voicemails are short and less than 30 seconds. How does the billing work?
We offer virtually unlimited transcription plan on per mailbox basis. The billing is mostly based on the number of mailboxes in your account.
- Is it ok to join a few audio files as one and process as a single voicemail?
No. To achieve best accuracy, Nexiwave Speech Engine performs heavy adaptation to speaker accent, background noise and line condition. Merging audio files, even for the same speaker, and processing as one will have significant detrimental effect and will very likely cause low accuracy.
- What audio formats and languages are supported?
Nearly all audio formats are supported. US/Canada English is the supported language at this point.
- I don't want anyone to listen to my voicemails. Is there any human involved in any of the transcription process?
No human is involved in any of the transcription process. Nexiwave uses the most advanced speech technology to convert your audio to text. Your privacy is 100% guaranteed.
- Are my voicemails stored on Nexiwave?
Nexiwave does NOT store any of your voicemails or its transcript. See "Single Delivery" policy: upon the success delivery of the transcript via email or SMS message, any content related data is purged from Nexiwave system.
- Is Nexiwave HIPAA Compliant?
Yes, we are. Read more here.