customers
All customer stories
Top Voice AI companies are building with Assembly.
resources
Latest Release
Universal 3.5 Pro
Our new flagship speech-to-text model handles real-world audio the way it actually happens. Available for real-time and pre-recorded audio.
resources
Transcribe 100% of your calls with high accuracy, then add speaker labels, sentiment, entity detection, and PII redaction for QA and compliance—all from one API built for telephony audio at scale.
Thanks for calling—this is a recorded line. Can I get your account number?
Sure, it's [redacted].
Got it. I can waive that fee and have this resolved within 24 hours.
Universal-3 Pro is tuned for 8kHz call audio, with multichannel support so agent and customer are cleanly separated.
Redact names, card numbers, and account data in both text and audio to support PCI and GDPR workflows across your call archive.
Per-utterance sentiment, entity detection, and summaries turn raw recordings into QA scores, escalation flags, and coaching signals.
36%
improvement in close rate
“The new Universal-3.5 Pro speech model from AssemblyAI is best so far in terms of accuracy, latency, and language switching.”
80%
increase in customer satisfaction
“Assembly has saved us countless hours managing models, and provided exceptional accuracy.”
36%
improvement in close rate
“The new Universal-3.5 Pro speech model from AssemblyAI is best so far in terms of accuracy, latency, and language switching.”
80%
increase in customer satisfaction
“Assembly has saved us countless hours managing models, and provided exceptional accuracy.”
Submit recordings in batch and get back transcripts with speaker separation, redacted PII, sentiment, and entities in a single response. Use webhooks for high-volume workloads so your QA system processes thousands of calls without polling.
Start building freeNo credit card required
Speech analytics is only as trustworthy as the transcript under it. Word error rate measures accuracy against a human reference on pre-recorded audio—the QA workload most analytics runs on.
Pre-recorded word error rate on English audio.
*Lower is better*Source: AssemblyAI published benchmarks — assemblyai.com/benchmarks.
