Recording analysis · Paid plans

AI sales call analysis for the recordings you upload

Upload a short call recording. Vozah transcribes it, scores the rep on the same nine dimensions used for AI practice calls, and helps you pick what to rehearse next.

Already on a paid plan? Sign in to upload a recording

How it works

From recording to next practice task

  1. 1. Export and upload

    Export audio from Zoom, Teams, Meet or your dialer as MP3, MP4, MPEG, MPGA, M4A, WAV or WebM, up to 4 MiB. Trim long calls to the conversation you want reviewed.

  2. 2. Transcription

    The audio goes to OpenAI's speech-to-text service. The result is a plain-text transcript without speaker labels. Uploads that produce no usable speech fail and do not use your allowance.

  3. 3. Nine-dimension scoring

    The scoring model infers which speaker is the rep, scores each of the nine dimensions from 0 to 100 with written feedback, and adds a coaching summary. The overall score is a weighted composite.

  4. 4. Choose the next drill

    Your scorecard joins your practice history. From the dashboard, a coaching plan finds your three weakest dimensions across recent scorecards and suggests matching practice scenarios.

Sample analysis

What a scored upload looks like

This is a hypothetical example written by Vozah to show the format. It is not a customer call. The rep, company and scores are invented. The dimension names, weights and composite formula match the product.

Transcript excerpt · Hypothetical

Uploaded transcripts arrive as unlabeled text, like this. The scorer decides who the rep is.

“Thanks for making time. I saw you're hiring three more field technicians, so I wanted to ask how you're handling scheduling today.”

“Mostly a shared calendar and a lot of phone calls. It works, mostly.”

“Got it. Our platform automates dispatch, sends reminders, syncs with QuickBooks and has a mobile app for technicians, so it would save you a lot of time.”

“We looked at something similar last year.”

“Totally fair. Why don't I send over some info and follow up next week?”

Uploaded call · Discovery · 9 minIllustrative example

57/100

Weighted composite of the nine dimension scores

  • Opening Hook70

    Specific reason for the call tied to their hiring news.

  • Discovery58

    Accepted “it works, mostly” without asking what breaks or what it costs.

  • Qualification45
  • Value Proposition62

    Listed features before a pain was confirmed.

  • Objection Handling66

    Did not ask what happened with last year’s evaluation.

  • Talk-to-Listen Ratio50
  • Pacing72
  • Closing40
  • Next-Step Clarity38

    “Follow up next week” has no time or owner.

Coaching summary: Good opener, but discovery stopped at the first answer and the call ended without a dated next step. Before the next call, practice asking two follow-up questions about the scheduling pain and proposing a specific meeting time.

Hypothetical example for illustration. Real scorecards show feedback for every dimension; some rows are shortened here.

How this example becomes practice

The three weakest dimensions here are Next-Step Clarity, Closing and Qualification. When the rep generates a coaching plan, Vozah averages recent scorecards, picks the weakest three and suggests a scenario for each. In this case that could be closing practice against the built-in “Procurement Pushback at the Close” buyer. The rep then runs a voice roleplay and compares the new scorecard with the uploaded call.

Limits

File size, length and monthly allowances

How much audio fits in 4 MiB

Approximate, at a constant bitrate. Container overhead and variable bitrates shorten these slightly.

FormatFits
MP3 or M4A, 32 kbps mono (speech export)about 17.5 min
MP3 or M4A, 64 kbpsabout 8.7 min
MP3, 128 kbps (common music default)about 4.4 min
WAV, 16 kHz 16-bit mono (256 kbps)about 2.2 min
WAV, 44.1 kHz 16-bit stereo (1,411 kbps)about 24 sec

Uploads per month

Per covered seat, reset on the first day of each UTC calendar month.

PlanPriceUploads
Free$0Not included
Solo$29/mo5
Team$149/mo50
Growth$399/mo200
Business$899/mo1,000
  • Up to 10 uploads per hour per user.
  • Very long transcripts are rejected. Trim the recording and upload again.
  • If scoring fails after transcription, the transcript is kept and you can retry scoring from your dashboard.
What it can and cannot tell you

Transcript-based, not acoustic

Useful for judging

  • Whether the opener gave a specific reason for the call
  • Whether discovery questions went past the first answer
  • How objections were acknowledged and explored
  • Whether value was tied to a pain the buyer stated
  • Whether the call ended with a dated, owned next step

Limits to know

  • No speaker labels: the rep is inferred and can be misidentified
  • Talk-to-listen balance is estimated from text, not timed audio
  • Tone, pitch and emotion in the audio are not measured
  • No deal outcome, CRM or calendar data is used
  • Scores are AI coaching guidance and can be wrong. Review them before acting on them

For the broader coaching workflow, see AI sales coach. For how the rubric dimensions are defined, read how AI sales call scoring works. Weighing this against a conversation intelligence platform? Read Vozah vs Gong.

Recordings and consent

Upload only what you are allowed to share

Recording and consent rules depend on where call participants are located. Make sure you have permission to record and to process the recording with an AI service, and leave out sensitive customer details. Vozah saves the transcript and scorecard, not the audio file. Read how Vozah handles recordings and transcripts and the privacy policy.

FAQ

Common questions

What file types and sizes can I upload?⌄
MP3, MP4, MPEG, MPGA, M4A, WAV and WebM files up to 4 MiB (4,194,304 bytes). Compressed speech audio fits far more conversation than WAV. Export audio only from your meeting tool, or trim the recording to the part you want reviewed.
Which plans include recording uploads?⌄
Solo includes 5 uploads a month, Team 50 per covered seat, Growth 200 per covered seat and Business 1000 per covered seat. Free accounts cannot upload recordings. Allowances reset on the first day of each UTC calendar month, and uploads that fail transcription do not count.
Does Vozah know which speaker is the sales rep?⌄
Not reliably. The transcript is plain text without speaker labels, and the scoring model infers which speaker is the rep from the conversation. Attribution can be wrong, especially on calls with several participants, so treat the talk-to-listen score on uploads as an estimate.
Does the analysis measure tone of voice?⌄
No. Scoring reads the transcript text. It can judge what was said, such as whether a question dug past the first answer, but it does not measure pitch, volume or emotion in the audio.
Is the audio file stored?⌄
Vozah sends the audio to OpenAI for transcription and does not save it as an audio file. The transcript and scorecard are saved to your account. Only upload recordings you have permission to use, and avoid sensitive customer details.
How is this different from conversation intelligence software?⌄
Conversation intelligence platforms typically record and index a team's calls automatically. Vozah analyzes individual recordings you choose to upload and connects the result to AI practice calls. It does not join meetings, sync a call library or integrate with a CRM.
Start practicing

Start with practice, add recordings later

Free accounts include 10 voice practice minutes a month. Recording uploads start on the Solo plan.

10 practice minutes each month · No credit card required