I'm using Grok (xAI) transcription with 3CX v20 U 9 (Grok 4.3) and have noticed that phone numbers and other numeric sequences are being transcribed as fully spelled-out words rather than digits.
Example:
Instead of:
337 230 4538
The transcript shows:
three three seven two three zero four five three eight
This makes it harder to quickly copy and paste numbers from voicemails or call transcripts.
Looking at the xAI Speech-to-Text documentation, the API supports Inverse Text Normalization via the format=true parameter (together with a language code). When enabled, it converts spoken numbers, phone numbers, currencies, etc. into proper written/digit form.
Is there currently any setting in 3CX that enables this format option for Grok transcriptions?
If not, would it be possible to add support for it (or enable it by default) so that numbers appear in digit form?
Thanks in advance!
Example:
Instead of:
337 230 4538
The transcript shows:
three three seven two three zero four five three eight
This makes it harder to quickly copy and paste numbers from voicemails or call transcripts.
Looking at the xAI Speech-to-Text documentation, the API supports Inverse Text Normalization via the format=true parameter (together with a language code). When enabled, it converts spoken numbers, phone numbers, currencies, etc. into proper written/digit form.
Is there currently any setting in 3CX that enables this format option for Grok transcriptions?
If not, would it be possible to add support for it (or enable it by default) so that numbers appear in digit form?
Thanks in advance!
