Call Recording Transcription & AI Call Analysis

Turn sales and support call recordings into a speaker-labelled transcript and a ready-to-use call analysis. Paste links to your recordings (MP3, WAV, M4A, MP4…, or Google Drive and Dropbox file links) and get, for every call:

• 🗣️ Transcript with speaker labels (speaker diarization): who said what and when, with names and roles (agent, customer, supervisor…) when they're said in the call, plus SRT subtitles.
• 🧠 AI call analysis: summary, outcome, resolved or not, sentiment (overall and per speaker), topics, action items with owner and due date, objections and how they were handled, pain points, competitors mentioned and key moments with timestamps.
• ✅ QA scorecard: your own yes/no checklist ("Did the agent confirm the customer's email?") answered for every call, with the quote and time that prove it.
• 📊 Talk-time stats, counted from the timestamps: talk share, turns, words per minute, questions asked, longest monologue, interruptions and long pauses.
• 🔒 Optional redaction of emails, phone numbers (and other long digit strings such as account numbers), card numbers, US SSNs and IBANs.

No API key, no account to connect, no setup. It works for sales calls, customer support, interviews and meetings, in English and most major languages.

What can you do with it?

• Sales call analysis: see why deals stall: objections, competitor mentions, next steps and how each rep handles price questions.
• Call center QA: score every support call against your checklist instead of sampling 2% by hand.
• CRM notes on autopilot: send the summary, outcome and action items to your CRM with Make, Zapier, n8n or a webhook.
• Coaching: quality score with specific strengths and improvements per call, plus talk ratio and monologue length.
• Searchable archive: speaker-labelled transcripts of every recorded call or meeting, in JSON, CSV or Excel.

What recordings does it accept?

Input	Example	Supported
Direct audio or video file link	https://example.com/recordings/call-1042.mp3	✅ mp3, wav, m4a, aac, ogg, opus, webm, flac, amr, mp4, mov…
Phone system / cloud storage file URL	a public recording URL from your telephony provider, S3 or another bucket	✅ if it opens without logging in
Google Drive file	https://drive.google.com/file/d/…/view	✅ shared as "Anyone with the link"
Dropbox file	https://www.dropbox.com/scl/fi/…?dl=0	✅
Folders, or pages that need a login		❌ share each file's own link

Telephone-quality (8 kHz) recordings work fine. Recordings up to 4 hours are supported; set Max minutes per recording to cap long files.

How much does it cost?

Pay per minute of audio and per analysed call. No monthly fee.

Event	Price (FREE/BRONZE)	Lower on paid plans
Call minute transcribed (speaker labels, timestamps, stats, SRT)	$0.01 / started minute	down to $0.008
Call analysis (summary, sentiment, action items, scorecard…)	$0.02 / call	down to $0.018
Actor start	$0.00005 / run	—

A 10-minute sales call costs $0.12 with the full analysis, or $0.10 for the transcript only. Calls longer than 30 minutes pay one analysis per started 30 minutes (a 45-minute call: 2 × $0.02). Nothing is charged for a recording that fails, and if the analysis fails the transcript is still saved and only the transcript is charged.

Cap your spend: set a maximum cost per run. The transcript comes first: when the budget can't pay for both, a call is transcribed without the analysis.

Free plan: the AI runs on paid infrastructure, so free-plan runs include a 3-minute trial per run; upgrade to process full calls.

How to analyse a call recording

1. Click Try for free. The example input analyses a 2.5-minute sample sales call (a demo call recorded with AI voices) and answers two scorecard questions.
2. Paste your own recording links into Recording links, one per line.
3. Optional: pick the Call type, add QA scorecard questions, choose the Analysis language, and set the Number of speakers if you know it (2 for most calls).
4. Run, then open the Calls, Call analysis or Transcripts view, or download JSON, CSV or Excel.

Input example

{
  "recordingUrls": ["https://example.com/recordings/call-1042.mp3"],
  "callType": "sales",
  "scorecardQuestions": ["Did the agent confirm the customer's contact details?", "Was a clear next step agreed?"],
  "analysisLanguage": "en",
  "expectedSpeakers": 2,
  "redactPII": false
}

Ready-made tasks

Open one, replace the sample recording with your own links, and run:

• Sales Call Analysis: Objections, Next Steps & Score
• Customer Support Call QA Scorecard
• Call Recording to Text with Speaker Labels
• Redacted Call Transcripts (Mask Emails, Phones, Cards)

Output

One record per recording. Real output from the example input (shortened):

{
 "url": "https://api.apify.com/v2/key-value-stores/T5HtoeQYut1ufwkKC/records/sample-sales-call.mp3",
 "durationSeconds": 147.7,
 "truncated": false,
 "language": "en-US",
 "speakerCount": 2,
 "speakers": [
  {"label": "S1", "name": "Sarah", "role": "agent", "talkSeconds": 76.0, "talkShare": 0.58, "turns": 10, "words": 219,
   "wordsPerMinute": 173, "questionsAsked": 5, "longestMonologueSeconds": 12.6},
  {"label": "S2", "name": "Mike Turner", "role": "customer", "talkSeconds": 55.0, "talkShare": 0.42, "turns": 9, "words": 139,
   "wordsPerMinute": 152, "questionsAsked": 3, "longestMonologueSeconds": 8.8}
 ],
 "stats": {"talkSeconds": 131.0, "silenceSeconds": 0.0, "longPauses": 0, "interruptions": 0, "speakerChanges": 18, "speechShare": 0.887},
 "summary": "Mike Turner called Brightline Solar to discuss a quote for a rooftop solar system. He expressed concerns about the price compared to a competitor, leading the agent to explain the value proposition and offer a promotional discount. The call concluded with the agent agreeing to send a revised quote and scheduling a site visit for the following Tuesday.",
 "outcome": "Site visit booked for Tuesday at 10:00 AM and a revised quote requested.",
 "sentiment": "positive",
 "qualityScore": 9.0,
 "analysis": {
  "callType": "sales",
  "resolved": true,
  "topics": ["solar system quote", "pricing", "warranty", "financing", "installation timeline", "site visit"],
  "actionItems": [
   {"owner": "Sarah", "task": "Email a revised quote with the promotion and financing option", "due": "Friday"},
   {"owner": "Sarah", "task": "Conduct site visit", "due": "Tuesday at 10:00 AM"}
  ],
  "objections": [
   {"objection": "Price is higher than competitor Sunpeak.", "response": "Highlighted premium panels, 25-year warranty, and 10-year labor coverage.", "resolved": true}
  ],
  "competitors": ["Sunpeak"],
  "keyMoments": [{"at": 36.1, "label": "Customer raises price objection"}, {"at": 65.2, "label": "Agent offers promotion"}, "…"],
  "strengths": ["Clear explanation of value proposition", "Proactive offer of a promotion"],
  "improvements": ["Could have asked about the customer's budget earlier in the conversation"],
  "scorecard": [
   {"question": "Did the agent confirm the customer's contact details?", "answer": "yes",
    "evidence": "Can you confirm the email address on the quote?", "at": 15.0},
   {"question": "Did the agent ask about the customer's budget?", "answer": "no", "evidence": null, "at": null}
  ]
 },
 "text": "[00:00] Sarah: Hi, thanks for calling Brightline Solar. This is Sarah. How can I help you today?\n[00:05] Mike Turner: Hey Sarah, my name is Mike Turner. I got a quote from you guys last week…",
 "segments": [{"start": 0.6, "end": 4.6, "speaker": "S1", "text": "Hi, thanks for calling Brightline Solar. This is Sarah. How can I help you today?"}, "…"],
 "srt": "1\n00:00:00,600 --> 00:00:04,600\nSarah: Hi, thanks for calling Brightline Solar…"
}

Times (at, start, end) are seconds from the start of the recording. Everything that must be exact (times, talk seconds, word and question counts, interruptions) is counted by code from the timestamps; the AI only writes the descriptions.

How fast is it?

Measured on Apify with the default 512 MB memory: a 2.5-minute call is transcribed and analysed in about 15 seconds, and a 30-minute recording in under 3 minutes. Recordings in one run are processed one after another.

How accurate is it?

Measured on test calls with a known script, in telephone quality (8 kHz):

• Speaker labels: 19 of 19 turns right on a two-person sales call (in each of 3 runs), and 14 or 15 of 15 on a three-person support call where a supervisor joins.
• Timestamps: median error 0.1 to 0.2 seconds.
• Best for 2–4 speakers. On a 9-speaker court hearing where several similar voices take turns, voices were merged into fewer speakers. For panels and large meetings, treat speaker labels as approximate.

Long recordings are processed in about 10-minute parts cut at pauses, and each part is told which speakers were already found, so a voice keeps its label for the whole call.

Use it with AI agents (MCP)

https://mcp.apify.com?tools=bossless/call-transcription-analyzer

Ask your assistant to "transcribe and analyse this call recording", and it runs this actor and reads back the summary and action items.

Use the API from code

curl -X POST "https://api.apify.com/v2/acts/bossless~call-transcription-analyzer/run-sync-get-dataset-items?token=YOUR_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"recordingUrls": ["https://example.com/recordings/call-1042.mp3"], "callType": "support"}'

from apify_client import ApifyClient

client = ApifyClient("YOUR_TOKEN")
run = client.actor("bossless/call-transcription-analyzer").call(run_input={
    "recordingUrls": ["https://example.com/recordings/call-1042.mp3"],
    "scorecardQuestions": ["Did the agent greet the customer by name?"],
})
for call in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(call["summary"], call["analysis"]["actionItems"])

To analyse every new recording automatically, start a run from your phone system's "recording ready" webhook, or from Make, Zapier or n8n, and send the results on with an Apify integration.

FAQ

Do I need an OpenAI, Deepgram or other API key? No. Transcription and analysis are included in the price.

Which languages work? The transcript is written in the language spoken; most major languages work. Choose the analysis language separately (English by default, or the call's own language).

What happens with a private or broken link? It's skipped with a clear message and not charged. A run where every link was wrong still succeeds, with a message saying what to check.

Is my data stored or used for training? Recordings are downloaded only for the run and deleted when it ends. Results are saved in your own Apify dataset. The AI model runs on Google Cloud (Vertex AI). Google's terms say it won't use this data to train or fine-tune its models; it may cache requests for up to 24 hours and log them for abuse monitoring.

How are minutes counted? Per started minute of audio actually transcribed, so a 61-second call is 2 minutes.

Can it tell agent from customer? Yes: each speaker gets a role from the conversation (agent, customer, supervisor…), and a name when it's said in the call. Setting Number of speakers helps when you know it.

Is this an official product of any phone or CRM provider? No. This is an independent tool and is not affiliated with or endorsed by any phone system, CRM or cloud storage provider.

Related tools

• Speaker Diarization & Transcription: who spoke when for any audio or video file, with RTTM, SRT and VTT.
• Chinese Video Transcriber: Bilibili videos and any audio/video link to timestamped text, translation and SRT.