whatmodel.sh · find the best speech-to-text models you can run

Speech-to-text models in the LibriSpeech test-clean WER model set

A leaderboard of the best speech-to-text models. Models are ranked by one score that combines 86 published benchmarks. Filter by your computer or budget to find the best one you can use.

0 models · scores combine 86 benchmarks · updated how it works · find a leaderboard · benchmark your machine
task[all tasks][long-form transcription][multilingual transcription]
Filter by your hardware and budget
+ more · LibriSpeech test-clean WER
set[all sets][Open ASR English short-form WER (March 2026) · 176][LibriSpeech test-clean WER · 2]
score[overall][within selected set]

Filter the model list while keeping overall scores. Models without a shared score scale are not yet calibrated; select a set to see its own scores.

Detect your exact chip or graphics card:

curl -G whatmodel.sh/stt -d max=0 --data-urlencode "mac=$(sysctl -n hw.memsize machdep.cpu.brand_string)"Mac

curl -G whatmodel.sh/stt -d max=0 --data-urlencode "ram=$(awk '/MemT/{print $2}' /proc/meminfo)" --data-urlencode "nv=$(nvidia-smi --query-gpu=name,memory.total --format=csv,noheader 2>/dev/null)"Linux

curl.exe -G whatmodel.sh/stt -d max=0 --data-urlencode "ram=$((gcim Win32_ComputerSystem).TotalPhysicalMemory)" --data-urlencode "nv=$(nvidia-smi --query-gpu=name,memory.total --format=csv,noheader 2>$null)"Windows

$ terminal

detecting your OS…

Run curl "whatmodel.sh/stt?set=librispeech-test-clean-wer%40speech-nvidia-dle-asr"to get this list in your terminal (add &format=json for agents).

ram[8][16][24][32][48][64][96][128][192][256]GB
gpu[apple silicon][graphics card][none]
price[local][$0.001][$0.002][$0.005][$0.01][any]max per audio min

detecting your OS…

#modelscore↓where↓speed↓RAM↓get itsources
#modelscore↓where↓speed↓RAM↓get itsources

No models match these filters.

0 of 180 catalogue models listed here · sources and licences

Models outside this leaderboard (2 models)

No current run option verified: Jasper 10x5 (NVIDIA DLE, A100 recipe) · QuartzNet 15x5 (NVIDIA DLE, 21.07 recipe)

* new or thinly tested: ranked on too few benchmark results · † only the maker’s own numbers so far · ~RAM is estimated; full context may need more memory