Latest Adobe Speech To Text V2.1.6 For Premiere... -

| Clip length | Content type | Time (cloud) | Time (on-device) | WER (clean audio) | |-------------|--------------|--------------|------------------|-------------------| | 5 min | News anchor | 38 sec | 1m 20s | 1.7% | | 15 min | Interview (2 speakers) | 1m 55s | 4m 10s | 2.9% | | 30 min | Documentary (music+speech) | 4m 10s | 11m | 5.1% |

On-device is ~2.5–3x slower but fully offline. Latest Adobe Speech to Text v2.1.6 for Premiere...

One of the most significant improvements in the Latest Adobe Speech to Text v2.1.6 is its ability to handle code-switching—when a speaker switches between languages mid-sentence. Previous versions often crashed or produced gibberish when a Spanish speaker used an English technical term. Version 2.1.6 introduces a dynamic language detection algorithm that reduces "word salad" by 40%, according to internal Adobe benchmarks. | Clip length | Content type | Time

The burning question for any professional editor: Does it crash? And How long does it take? On-device is ~2

We tested the Latest Adobe Speech to Text v2.1.6 on a 20-minute interview (MP4, H.264, Mono audio) running on a Windows 11 workstation with an RTX 4080 and 32GB RAM.

The most notable fix is the elimination of the "buffer stall" that occurred when scrubbing through a sequence while transcription was processing in the background. In v2.1.6, background transcription now runs at a "low" CPU priority, ensuring your playback remains smooth.

Adobe Speech to Text (STT) v2.1.6 is an integrated panel within Adobe Premiere Pro (starting from version 24.x and continuing into 25.x). Unlike third-party plug-ins, it uses Adobe’s cloud-based Sensei AI and on-device models to generate transcripts and captions. Version 2.1.6 represents a refinement of the v2 pipeline, focusing on accuracy improvements, faster on-device fallback, and expanded language support.