---
title: "AssemblyAI"
type: "AI Tool"
url: "https://aidemos.com/tools/assemblyai-speech-to-text"
description: "Three batch clips in, AssemblyAI returned JSON with word timestamps, confidence, and speaker labels; overlap-heavy meetings still dropped 1,976 words."
category: "audio-speech"
website: "https://www.assemblyai.com/docs"
published: "2026-08-13T09:32:16.283855+00:00"
updated: "2026-09-01T02:57:17.687529+00:00"
evidenceCount: 22
verifiedCount: 19
coverage: "dense"
---

# AssemblyAI

Fast batch speech-to-text with rich metadata, strong jargon and mixed-language results, but overlap-heavy meetings can still lose too much.

## TL;DR Verdict

**Strong batch STT, but crosstalk is the weak spot**

**Where it wins:**
- You need batch transcription that returns completed JSON with word-level timestamps, confidence values, and speaker labels.
- You transcribe technical narration or jargon-heavy audio and want low WER on those clips.
- You need mixed-language batch transcription and can accept that this benchmark only covered a mostly-English code-switching sample.

**Main limitation:** You need overlap-heavy meeting audio to retain every word.

**Pricing:** Free tier $50 in free credits · Pay-as-you-go — Universal-3.5 Pro $0.21 / hr · Pay-as-you-go — Universal-2 $0.15 / hr · Streaming — Universal-3.5 Pro Realtime $0.45 / hr

`Batch API` · `Word timestamps` · `Speaker labels` · `0.016–0.018 RTF`

**Website:** [Visit AssemblyAI](https://www.assemblyai.com/docs)

## Evidence (first-party, tested)

*22 tested cells · 19/22 artifact-verified. Scores are out of 5. Cite a cell by its Evidence ID, e.g. `ev:assemblyai-speech-to-text·cross·automation-level`.*

| Criterion | Scenario | Verdict | Score | Proof | Evidence ID |
| --- | --- | --- | --- | --- | --- |
| Automation level | cross-scenario | ✓ worked | — | 🧾 [proof](https://d3epheqghktydj.cloudfront.net/research-media-assemblyai-da842a17ac10.md) | `ev:assemblyai-speech-to-text·cross·automation-level` |
| Automation level | Bilingual Spanish-English code-switching speech | ✓ worked | — | 🧾 [proof](https://cdn.futuresmart.ai/public/aidemos/d7fafcb3ae3242b1921a213ef20b723c.png?v=1) | `ev:assemblyai-speech-to-text·bilingual-spanish-english-code-switching-speech·automation-level` |
| Automation level | Overlapping meeting speech with cross-talk | ✓ worked | — | 🧾 [proof](https://cdn.futuresmart.ai/public/aidemos/feee65d2155c4e6db05a12c93df7cca4.png?v=1) | `ev:assemblyai-speech-to-text·overlapping-meeting-speech-with-cross-talk·automation-level` |
| Automation level | Medical anatomy narration with dense jargon | ✓ worked | — | 🧾 [proof](https://cdn.futuresmart.ai/public/aidemos/ae94ab5021b14f908a207a6926c5b793.png?v=1) | `ev:assemblyai-speech-to-text·medical-anatomy-narration-with-dense-jargon·automation-level` |
| Export | cross-scenario | ✓ worked | — | 🧾 [proof](https://cdn.futuresmart.ai/public/aidemos/9516365adcba407ab3445920d34b0f17.png?v=1) | `ev:assemblyai-speech-to-text·cross·export` |
| Export | Overlapping meeting speech with cross-talk | ✓ worked | — | 🧾 [proof](https://cdn.futuresmart.ai/public/aidemos/9f0a3e74b1f24fe98c2268f378a30952.png?v=1) | `ev:assemblyai-speech-to-text·overlapping-meeting-speech-with-cross-talk·export` |
| Export | Medical anatomy narration with dense jargon | ✓ worked | — | 🧾 [proof](https://cdn.futuresmart.ai/public/aidemos/81cd78fe36a94344bfa11676cf2dc1e3.png?v=1) | `ev:assemblyai-speech-to-text·medical-anatomy-narration-with-dense-jargon·export` |
| Export | Bilingual Spanish-English code-switching speech | ✓ worked | — | 🧾 [proof](https://cdn.futuresmart.ai/public/aidemos/3461c7c658d945cfab85abeade0b44b8.png?v=1) | `ev:assemblyai-speech-to-text·bilingual-spanish-english-code-switching-speech·export` |
| Input handling | Overlapping meeting speech with cross-talk | ✓ worked | — | 🧾 [proof](https://cdn.futuresmart.ai/public/aidemos/0399e932b1c14c8589fd34fd8a67f5ff.png?v=1) | `ev:assemblyai-speech-to-text·overlapping-meeting-speech-with-cross-talk·input-handling` |
| Input handling | cross-scenario | ✓ worked | — | 🧾 [proof](https://d3epheqghktydj.cloudfront.net/research-media-assemblyai-da842a17ac10.md) | `ev:assemblyai-speech-to-text·cross·input-handling` |
| Input handling | Bilingual Spanish-English code-switching speech | ✓ worked | — | 🧾 [proof](https://cdn.futuresmart.ai/public/aidemos/0ce99e4559074127a70999bb7a341dcd.png?v=1) | `ev:assemblyai-speech-to-text·bilingual-spanish-english-code-switching-speech·input-handling` |
| Input handling | Medical anatomy narration with dense jargon | ✓ worked | — | 🧾 [proof](https://cdn.futuresmart.ai/public/aidemos/c3c3285e919246b3aa0ad945156259f7.png?v=1) | `ev:assemblyai-speech-to-text·medical-anatomy-narration-with-dense-jargon·input-handling` |
| Input handling | Overlapping meeting speech and cross-talk | ✓ worked | — | 🧾 [proof](https://cdn.futuresmart.ai/public/aidemos/0399e932b1c14c8589fd34fd8a67f5ff.png?v=1) | `ev:assemblyai-speech-to-text·overlapping-meeting-speech-and-cross-talk·input-handling` |
| Input handling | Medical anatomy dictation with dense jargon | ✓ worked | — | 🧾 [proof](https://cdn.futuresmart.ai/public/aidemos/c3c3285e919246b3aa0ad945156259f7.png?v=1) | `ev:assemblyai-speech-to-text·medical-anatomy-dictation-with-dense-jargon·input-handling` |
| Input handling | Spanish-English code-switching conversation | ✓ worked | — | 🧾 [proof](https://cdn.futuresmart.ai/public/aidemos/0ce99e4559074127a70999bb7a341dcd.png?v=1) | `ev:assemblyai-speech-to-text·spanish-english-code-switching-conversation·input-handling` |
| Output quality | cross-scenario | ◐ mixed | — | 🧾 [proof](https://d3epheqghktydj.cloudfront.net/research-media-assemblyai-da842a17ac10.md) | `ev:assemblyai-speech-to-text·cross·output-quality` |
| Output quality | Bilingual Spanish-English code-switching speech | ◐ mixed | 21.0/5 | 🧾 [proof](https://cdn.futuresmart.ai/public/aidemos/0ce99e4559074127a70999bb7a341dcd.png?v=1) | `ev:assemblyai-speech-to-text·bilingual-spanish-english-code-switching-speech·output-quality` |
| Output quality | Medical anatomy narration with dense jargon | ✓ worked | 3.8/5 | 🧾 [proof](https://cdn.futuresmart.ai/public/aidemos/c3c3285e919246b3aa0ad945156259f7.png?v=1) | `ev:assemblyai-speech-to-text·medical-anatomy-narration-with-dense-jargon·output-quality` |
| Output quality | Overlapping meeting speech with cross-talk | ⚠ struggled | 33.2/5 | 🧾 [proof](https://cdn.futuresmart.ai/public/aidemos/0399e932b1c14c8589fd34fd8a67f5ff.png?v=1) | `ev:assemblyai-speech-to-text·overlapping-meeting-speech-with-cross-talk·output-quality` |
| Output quality | Overlapping meeting speech and cross-talk | ⚠ struggled | 33.2/5 | 🧾 [proof](https://cdn.futuresmart.ai/public/aidemos/b4efedf9cf6243739c92905b2983295f.png?v=1) | `ev:assemblyai-speech-to-text·overlapping-meeting-speech-and-cross-talk·output-quality` |
| Output quality | Medical anatomy dictation with dense jargon | ✓ worked | 3.8/5 | 🧾 [proof](https://cdn.futuresmart.ai/public/aidemos/b157a95118d242a09aa47af6cadc2a6b.png?v=1) | `ev:assemblyai-speech-to-text·medical-anatomy-dictation-with-dense-jargon·output-quality` |
| Output quality | Spanish-English code-switching conversation | ◐ mixed | 21.0/5 | 🧾 [proof](https://cdn.futuresmart.ai/public/aidemos/8b1086623b904d06bf9df2957e53e346.png?v=1) | `ev:assemblyai-speech-to-text·spanish-english-code-switching-conversation·output-quality` |

> 🧾 = artifact-verified (proof captured) · 👁 = observed (noted, no artifact) · verdicts: worked / mixed / struggled / failed.

## Observed performance across the three benchmark clips

Best on medical jargon, strongest on the bilingual clip, weakest on crosstalk.

- **3/3** Overlapping Speech / Crosstalk — 33.16% WER; 1,976 deletions on the four-way overlap meeting.
- **1/3** Medical Jargon — 3.78% WER and 100% recall of the scored jargon terms.
- **2/3** Bilingual Code-Switching — 21.04% WER and 72.5% Spanish token recall on a mostly-English clip.

> **Strong batch STT, but crosstalk is the weak spot**
>
> AssemblyAI completed all three batch jobs and stayed consistently fast, returning completed JSON with word-level timestamps, confidence, and speaker labels every time. It was excellent on the medical-jargon clip and best on the bilingual code-switching clip, but the overlapping-speech sample dropped 1,976 words, so it is a solid choice for structured batch transcription as long as overlap-heavy meetings are not the main workload.

## Demo Recording

[Video: AssemblyAI demo recording](https://cdn.futuresmart.ai/public/aidemos/76bbff2460ce4ea982ee151c85ee473d.mov?v=1)
*Video — Tutorial recording of the AssemblyAI batch transcription benchmark run.*

## Feature-by-Feature Breakdown

### Robust Speech-to-Text Transcription — 33.16/100

**Verdict:** Too much content is lost on crosstalk-heavy audio.

Transcribes difficult speech inputs into text, including overlap-heavy conversations, dense technical narration, and Spanish-English code-switching. The evidence came from the overlap-heavy meeting clip, the Gray's Anatomy clip, and a mostly-English mixed-language sample.

**Input:**

> **Audio**

**Output:**

**Input:**

> **Audio**

**Output:**

**Input:**

> **Audio**

**Output:**

**Bottom line:** It can process overlap-heavy meetings, but the omission rate is high enough that this is not a safe transcript-of-record result.

### Transcript Metadata Export

**Verdict:** Consistently available on every run.

Returns structured transcript payloads with word-level timestamps, confidence values, and related JSON fields. The evidence shows these metadata fields present consistently across the tested audio runs.

**Input:**

> **Audio**

**Output:**

**Input:**

> **Audio**

**Output:**

**Input:**

> **Audio**

**Output:**

**Bottom line:** This is a dependable structured-transcript export: timestamps, confidence, and deep JSON payloads were present on every run.

### Speaker Diarization

**Verdict:** Labels are detected, but attribution correctness was not measured.

Adds speaker labels or speaker IDs to transcript output and reports the number of distinct speakers detected. It was exercised on crosstalk, medical-jargon, and bilingual clips, where speaker labels were surfaced consistently.

**Input:**

> **Audio**

**Output:**

**Input:**

> **Audio**

**Output:**

**Input:**

> **Audio**

**Output:**

**Bottom line:** Speaker labels are exposed consistently, but this benchmark did not verify whether every attribution was correct.

### Batch Transcription

**Verdict:** Completed successfully on all three test runs.

Accepts uploaded audio files, runs them through a batch transcription flow, and returns a completed transcript JSON response. The evidence covers overlapping speech, medical narration, and bilingual code-switching inputs reaching completion through the API path.

**Input:**

> **Audio**

**Output:**

**Input:**

> **Audio**

**Output:**

**Input:**

> **Audio**

**Output:**

**Bottom line:** The batch API path was reliable across all three runs, with no manual intervention needed to reach completed JSON.

## Official pricing

Benchmark runs used Universal-3.5 Pro; diarization stacks on top as an add-on.

| Plan | Price | Notes |
| --- | --- | --- |
| Free tier | $50 in free credits | No credit card required; 5 new streaming connections/min. |
| Pay-as-you-go — Universal-3.5 Pro ★ (tested) | $0.21 / hr | 100 new streams/min; most accurate async model; 18 languages, native code switching. |
| Pay-as-you-go — Universal-2 | $0.15 / hr | 99 languages; "exceptional accuracy at a lower price." |
| Streaming — Universal-3.5 Pro Realtime | $0.45 / hr | Billed on WebSocket session duration, not audio duration. |

*The report says add-ons stack on top of the base rate, and multichannel audio is billed per channel.*

## Is It Right For You?

**Use it if**
- You need batch transcription that returns completed JSON with word-level timestamps, confidence values, and speaker labels.
- You transcribe technical narration or jargon-heavy audio and want low WER on those clips.
- You need mixed-language batch transcription and can accept that this benchmark only covered a mostly-English code-switching sample.

**Skip it if**
- You need overlap-heavy meeting audio to retain every word.
- You need validated speaker-attribution correctness rather than detected speaker labels.
- You need streaming or live latency from this evaluation; it was not measured.
- You need a balanced bilingual routing test; input-3 was mostly English.

## Classification

- **Category:** audio-speech
- **Subcategory:** audio-to-text
- **Type:** text
- **Built for:** Other

## Frequently Asked Questions

**Q: How accurate was AssemblyAI on overlapping speech?**

On the overlapping-speech / crosstalk clip, it scored 33.16% WER with 461 substitutions, 1,976 deletions, and 76 insertions against a 7,579-word reference. The run completed and detected 4 speaker labels, but too much content was omitted for transcript-of-record use.

**Q: How did it perform on medical jargon?**

It performed very well on the medical-jargon clip: 3.78% WER, 71 substitutions, 14 deletions, and 18 insertions against a 2,728-word reference. The transcript detail also reported 100% recall of the scored jargon terms.

**Q: Can it handle Spanish-English code-switching?**

Yes. On the bilingual code-switching clip it scored 21.04% WER and 72.5% Spanish token recall, and it returned a completed transcript without special configuration. The report also notes that the sample was mostly English, so balanced bilingual routing was not directly stress-tested.

**Q: What metadata did the JSON response include?**

The raw responses showed word-level timestamps, confidence values, and speaker labels. The payload depth was reported as 3/3, with deep JSON nesting and distinct speaker keys present in every run.

**Q: Was streaming tested in this benchmark?**

No. This was a batch benchmark, and streaming latency was explicitly not measured.

**Q: What pricing did the report list?**

The report lists a $50 free tier in credits, Universal-3.5 Pro at $0.21/hr, Universal-2 at $0.15/hr, and Universal-3.5 Pro Realtime at $0.45/hr. It also says add-ons stack separately, including async diarization at +$0.02/hr.

## Similar Tools

AI tools similar to AssemblyAI:

- [Speechmatics](https://aidemos.com/tools/speechmatics) — Strong batch STT for hard English audio, but weak on code-switching as configured.
- [OpenAI](https://aidemos.com/tools/openai) — Batch speech-to-text with word timestamps, but a strict upload cap and weak multilingual performance make it a mixed fit for hard audio.
- [AWS Transcribe](https://aidemos.com/tools/aws-transcribe) — Batch speech-to-text with timestamps and speaker labels, but weak on multilingual audio.
- [ElevenLabs Scribe](https://aidemos.com/tools/elevenlabs-scribe) — Fast batch speech-to-text with word-level metadata, strongest on jargon and weaker on overlap/code-switching.
- [Rev AI](https://aidemos.com/tools/rev-ai) — Low-cost batch speech-to-text with structured JSON, word timestamps, and speaker labels, but mixed accuracy on hard audio and weak multilingual recall.
- [Gladia](https://aidemos.com/tools/gladia) — Batch STT with rich JSON metadata and strong jargon recall, but weak overlap handling and a channel-duplication caveat on bilingual audio.
- [Deepgram](https://aidemos.com/tools/deepgram) — Batch speech-to-text with rich metadata, but weak on crosstalk and code-switching.
- [ElevenLabs](https://aidemos.com/tools/elevenlabs) — Natural-sounding voice cloning and narration, but with only approximate voice identity.

## Need a custom AI solution for this use case?

If you are looking to build a custom speech-to-text, transcription, or meeting transcription system for your business or internal workflow, email us at [contact@futuresmart.ai](mailto:contact@futuresmart.ai).

### Found something inaccurate or missing?

We try to keep our AI research accurate and useful. If you found outdated information, an issue, or have a suggestion, email us at [collaborate@aidemos.com](mailto:collaborate@aidemos.com).