XoomVoice turns hotel call recordings into transcripts, summaries and sentiment your team can act on — without storing the audio permanently.
Guest calls are full of feedback and operational signals, but they're hard to review at scale. XoomVoice closes the gap between a long call and a fast decision.
Listening to every call is slow and inconsistent. Each recording becomes a full transcript plus a concise, factual summary your team can read in seconds.
Complaints, requests and service issues are easy to miss in a long call. Every conversation is classified as positive, neutral or negative, with a score to back it up.
Call recordings are unstructured and hard to route. A short, scan-friendly title and a structured result turn every call into a clean, consistent record.
Gemini hears the recording directly and returns everything in a single flow — no separate speech-to-text service, no audio conversion step.
Guests switch languages mid-sentence, so XoomVoice doesn't force them into one. The recording's language is detected automatically, and Hindi, English and mixed speech are preserved exactly as spoken — never translated.
Teams can pass an optional language hint for extra context, and each title is written in the call's dominant language.
Send a downloadable link to the call recording and get a job reference back immediately. XoomVoice securely downloads and analyses the audio in the background while your system checks one reference until the result is ready.
Structured results slot into the calling, CRM or dashboard workflows your team already uses.
Call recordings are sensitive, so XoomVoice keeps them only as long as it needs to. Temporary audio is deleted the moment a job succeeds, and signed links and access headers never appear in logs.
File-size and timeout limits protect every download, and failed attempts retry automatically — with a short-lived cache so a retry doesn't need a fresh download.
Calm, precise and dependable — a lightweight, API-first service built to turn recordings into insight reliably.
Transcription, title, summary and sentiment in one flow — no separate speech-to-text service.
Pending, Downloading, Analyzing, Completed or Failed — always know where a recording stands.
Every result is checked before completion, and an empty transcript is never returned as a success.
Failed attempts retry on the same job — three times by default — with one corrective model retry for invalid output.
A 100 MB default file limit and a 15-minute processing window keep every job in check.
Works with signed and gateway-protected links, while links and access headers stay out of logs.
Common audio types are recognised, and unknown files are identified from their signatures — no conversion step.
Concurrency, file limits, timeouts, retries, allowed origins and model choice are all tuned to your setup.
Clear answers on what XoomVoice returns, how it handles audio and how it fits your team.
Still curious? Talk to our teamFour things: a short, scan-friendly title; a full verbatim transcript in the call's original language; a factual summary in one to three sentences; and the overall sentiment — positive, neutral or negative — with a score from 0.0 to 1.0.
Yes. The language is detected automatically, and Hindi, English and mixed-language speech are transcribed as spoken rather than translated. You can also pass an optional language hint, and the title follows the call's dominant language.
No. Share a downloadable link to the recording — including signed links or links behind a protected recording gateway — and XoomVoice fetches the audio securely in the background.
Only temporarily. Audio is deleted as soon as a job succeeds. If an attempt fails, the audio may be held briefly in a short-lived local cache so a retry doesn't need a fresh download. Recording links and access headers are kept out of logs.
Failed attempts retry automatically on the same job — three times by default. If a recording still can't be analysed, you get a clear, safe failure message, such as an expired download link, without exposing links, credentials or file paths.
Results come back as structured data through a simple API, ready for the calling, CRM or dashboard workflows you already use. There's no upload form, no separate transcription service and no database to set up.