I am thrilled to announce a new online tool I’ve created called VoxHumana (VxH). With VoxHumana, you upload an audio file, and it’ll send it through Whisper for automatic transcription, then to the Montreal Forced Aligner (MFA) for forced alignment, and then through new-fave for automatic formant extraction. With VxH, you can take raw audio and get a spreadsheet of acoustic measurements.
VxH is designed to be a “spiritual successor” to DARLA. By that, I mean that it is clearly inspired by DARLA but is not a direct continuation of it. VxH does much of what DARLA could do: transcription, forced alignment, and formant extraction. But it uses the latest software for each one. Instead of an in-house transcription method, VxH uses Whisper. Instead of MFA version 1.X, VxH uses MFA version 2.x (with version 3.x coming soon). And instead of FAVE-Extract, it uses new-fave.
To be clear, there is very little that VxH does that these component parts don’t do already. There is no in-house processing other than what Whisper, MFA, and new-fave offer. VxH literally strings the three pieces of software together into an easy-to-use interface.