Recordings filed to the right place by rule
When recordings pour in from numbers, uploads, and SFTP, routing rules decide where each one belongs: which project, which team, which agent. By voice, by number, or by what was said.
Recording calls is the easy part. The quiet, expensive problem is sorting them: whose call is this, which team does it belong to, which project should score it? Do that by hand and it consumes hours; skip it and every downstream report is built on misfiled data.
IdentityCall routes each incoming recording automatically. Rules match on who is speaking (by voice, not caller ID), on the phone number, or on the transcript content, in that order of precedence. Matches below your threshold wait in a review queue for a human decision.
Why sorting recordings breaks down
Manual sorting does not scale
Someone drags recordings into the right project until volume grows, then the backlog does the filing and nothing is where it should be.
Caller ID only names the line
On shared desks and rotating shifts, the number tells you which phone was used, not which agent actually spoke. Attribution built on extensions is guesswork.
Misroutes are silent
A misfiled recording does not announce itself. It quietly skews an agent’s scorecard, a team’s pass rate, and every report built on top.
How routing works in IdentityCall
Voice-match rules
Each agent enrolls a voice profile. IdentityCall recognizes who is actually speaking on a recording, by acoustic fingerprint, and files the call to them, even on shared lines and borrowed desks.
Phone and transcript rules
Route by the number that was called or by what was said on the call. Rules apply in a clear order of precedence: voice first, then phone, then transcript.
Confidence you control
Every voice match carries a confidence score, and you set the bar it must clear. Below the threshold, the recording goes to the review queue instead.
Shadow mode before go-live
Run rules in dry-run mode first: IdentityCall logs what each rule would have done without moving anything, so you can compare against reality before switching routing on.
A review queue for the unsure
Matches that clear detection but not your threshold land in a review queue. A human resolves, keeps, or ignores each one.
Duplicate profiles, merged
When the same person ends up with two voice profiles, IdentityCall flags the likely duplicate with a similarity score and suggests a merge you approve or dismiss.
What you get
Recordings filed to the right project without manual sorting- Per-agent attribution by voice, even on shared numbers
- Matches below your threshold wait for review
- Rules tested in shadow mode before they act
- Voice profiles kept clean with merge suggestions
- Scorecards and reports built on correctly attributed calls
Frequently asked questions
How does voice-based routing work?
Agents enroll a short voice sample that becomes a biometric profile. When a recording arrives, IdentityCall compares the voices on it against enrolled profiles and routes the call to the matching agent, applying the confidence threshold you set.
What happens when the system is not sure?
Matches below your confidence bar go to a review queue where a person resolves them. A confident wrong match is still possible, so compare routing decisions against reality in shadow mode before going live.
Can I test rules before turning them on?
Yes. Shadow mode runs every rule against incoming recordings and records what would have happened, without routing anything. Review the dry-run decisions, tune the rules, then enable.
What if an agent has two voice profiles?
IdentityCall detects likely duplicates by comparing profiles and suggests merges with a similarity score. You approve or dismiss each suggestion; merging combines the profiles so attribution stays consistent.
Related
Let recordings file themselves
See how voice, number, and transcript rules put every call where it belongs.
