Search Speech Model Match
Every model profile, comparison, guide, language and glossary term on one page (197 entries). Type to filter, or browse by section.
Tools (6)
- Model matcherRank all reviewed models for your language, hardware and workload
- Interactive model comparisonTwo or three models side by side
- WER calculatorWord and character error rate
- Benchmark and capacity plannerCompare runs on your own audio
- Language support matrix104 languages
- Popularity trackerLive Hugging Face downloads, likes and trending ASR models
Models (26)
- Whisper tinyOpenAI · 39M · 99+ listed languages · MIT
- Whisper baseOpenAI · 74M · 99+ listed languages · MIT
- Whisper smallOpenAI · 244M · 99+ listed languages · MIT
- Whisper mediumOpenAI · 769M · 99+ listed languages · MIT
- Whisper turboOpenAI · 809M · 99+ listed languages · MIT
- Parakeet TDT 0.6B v3NVIDIA · 600M · 25 European languages · CC BY 4.0
- OrukeetOruk AI · 627M · 25 European languages · CC BY-SA 4.0
- Nemotron 3.5 ASR Streaming 0.6BNVIDIA · 600M · 40 language-locales in tiers · OpenMDW-1.1
- Qwen3-ASR 0.6BQwen · 0.6B · 30 languages + 22 Chinese dialects · Apache-2.0
- Qwen3-ASR 1.7BQwen · 1.7B · 30 languages + 22 Chinese dialects · Apache-2.0
- Distil-Whisper large-v3Hugging Face / Distil-Whisper · large-v3 distilled · English · MIT
- Moonshine Tiny Streaming (English)Moonshine AI · 34M · English · MIT
- Moonshine Medium Streaming (English)Moonshine AI · 245M · English · MIT
- Omnilingual ASR CTC 300MMeta AI · 325M · 1,600+ language codes · Apache-2.0
- Cohere Transcribe 03-2026Cohere Labs · 2B · 14 listed languages · Apache-2.0
- VibeVoice-ASR 7BMicrosoft · 9B total · 51 listed languages · MIT
- VibeVoice-ASR Streaming 1.5BMicrosoft · 3B total · 10 listed languages · MIT
- VibeVoice-ASR-BitNetMicrosoft · 0.3B quantized · 7 listed languages · MIT
- Fun-ASR-MLT-Nano-2512FunAudioLLM · 800M · 31 listed languages · Apache-2.0
- Voxtral Mini 4B Realtime 2602Mistral AI · 4B · 13 listed languages · Apache-2.0
- Granite Speech 5.0 2BIBM · 2B · 6 listed languages · Apache-2.0
- Granite Speech 5.0 470M TurboCTCIBM · 470M · 7 listed languages · Apache-2.0
- VibeVoice-ASR-Streaming-7BMicrosoft · 9B total · 10 listed languages · MIT
- Parakeet ReduxMoondream · 1.58-bit Parakeet 0.6B (178 MB) · 25 European languages · CC BY 4.0
- WhistleCactus Compute · 16.9 MB file (parameters not stated) · 7 listed languages · Apache-2.0
- Jais-ASRIFM · 5.7B total · Arabic (MSA + 19 dialects), plus English, French, Spanish, Italian · Apache-2.0
Comparisons (12)
- Whisper turbo vs Voxtral Mini 4B Realtime 2602Batch multilingual workhorse against a purpose-built realtime model.
- Whisper medium vs Whisper turboTwo large-ish Whisper checkpoints with very different speed profiles.
- Whisper turbo vs Distil-Whisper large-v3Multilingual turbo against the English-only distilled large-v3.
- Whisper turbo vs VibeVoice-ASR 7BGeneral transcription against long-form, speaker-attributed transcription.
- Parakeet TDT 0.6B v3 vs OrukeetNVIDIA's Parakeet against its Parakeet-derived community release.
- Nemotron 3.5 ASR Streaming 0.6B vs Voxtral Mini 4B Realtime 2602Two native streaming models with very different footprints.
- Whisper turbo vs Cohere Transcribe 03-2026The default open baseline against Cohere's 2B open-weights model.
- Qwen3-ASR 0.6B vs Fun-ASR-MLT-Nano-2512Two compact multilingual models with strong Chinese coverage.
- Parakeet TDT 0.6B v3 vs Parakeet ReduxNVIDIA's Parakeet against Moondream's 1.58-bit, CPU-first version of the same model.
- Moonshine Tiny Streaming (English) vs WhistleTwo tiny native-streaming models for phones and embedded devices.
- Whisper turbo vs Granite Speech 5.0 2BThe default open baseline against IBM's October 2026 Apache-2.0 model.
- VibeVoice-ASR Streaming 1.5B vs VibeVoice-ASR-Streaming-7BTwo sizes of Microsoft's native streaming who-said-what model.
New releases tracked outside the matcher (1)
Guides (20)
- Best local speech-to-text for meetings and speaker attribution
- Best local speech-to-text for Raspberry Pi, phones and other edge devices
- Best local speech-to-text for subtitles and timestamps
- Best local speech-to-text for voice agents and assistants
- Best local speech-to-text models for Apple Silicon
- Best local speech-to-text models for CPU-only PCs
- Best local speech-to-text models for multilingual use
- Best local speech-to-text models for streaming
- Best local speech-to-text models in 2026
- Best local STT models for an 8 GB GPU
- Best local Whisper alternatives in 2026
- How much memory do local speech-to-text models need?
- Moonshine vs Whisper: live voice or general-purpose ecosystem?
- New open speech-to-text models: autumn 2026 releases
- Omnilingual ASR vs Whisper: breadth or ecosystem?
- Parakeet v3 vs Nemotron 3.5 ASR Streaming
- Qwen3-ASR vs Whisper: which local ASR family fits better?
- Which local speech-to-text models can you use commercially?
- Whisper vs faster-whisper: same model family, different runtime
- Whisper vs Parakeet: which local ASR should you test first?
Language guides (23)
- Best local speech-to-text for English26 models list English
- Best local speech-to-text for Spanish21 models list Spanish
- Best local speech-to-text for French22 models list French
- Best local speech-to-text for Portuguese21 models list Portuguese
- Best local speech-to-text for Mandarin Chinese16 models list Mandarin Chinese
- Best local speech-to-text for Japanese17 models list Japanese
- Best local speech-to-text for Russian16 models list Russian
- Best local speech-to-text for Polish16 models list Polish
- Best local speech-to-text for Dutch16 models list Dutch
- Best local speech-to-text for Arabic14 models list Arabic
- Best local speech-to-text for Hindi12 models list Hindi
- Best local speech-to-text for Turkish9 models list Turkish
- Best local speech-to-text for Ukrainian11 models list Ukrainian
- Best local speech-to-text for Vietnamese12 models list Vietnamese
- Best local speech-to-text for Indonesian9 models list Indonesian
- Best local speech-to-text for Czech14 models list Czech
- Best local speech-to-text for Swedish13 models list Swedish
- Best local speech-to-text for Greek13 models list Greek
- Best local speech-to-text for Hebrew6 models list Hebrew
- Best local speech-to-text for Persian8 models list Persian
- Best local speech-to-text for Cantonese8 models list Cantonese
- Best local speech-to-text for German20 models list German
- Best local speech-to-text for Italian21 models list Italian
All languages (81)
- Korean16 models · code ko
- Danish13 models · code da
- Finnish13 models · code fi
- Hungarian13 models · code hu
- Romanian13 models · code ro
- Croatian12 models · code hr
- Slovak12 models · code sk
- Bulgarian11 models · code bg
- Estonian11 models · code et
- Slovenian11 models · code sl
- Latvian10 models · code lv
- Lithuanian10 models · code lt
- Filipino9 models · code fil
- Malay9 models · code ms
- Maltese9 models · code mt
- Thai9 models · code th
- Macedonian7 models · code mk
- Afrikaans6 models · code af
- Armenian6 models · code hy
- Catalan6 models · code ca
- Icelandic6 models · code is
- Javanese6 models · code jv
- Mongolian6 models · code mn
- Nepali6 models · code ne
- Norwegian6 models · code no
- Serbian6 models · code sr
- Swahili6 models · code sw
- Urdu6 models · code ur
- Yiddish6 models · code yi
- Albanian5 models · code sq
- Amharic5 models · code am
- Assamese5 models · code as
- Azerbaijani5 models · code az
- Bangla5 models · code bn
- Bashkir5 models · code ba
- Basque5 models · code eu
- Belarusian5 models · code be
- Bosnian5 models · code bs
- Breton5 models · code br
- Burmese5 models · code my
- Faroese5 models · code fo
- Galician5 models · code gl
- Georgian5 models · code ka
- Gujarati5 models · code gu
- Haitian Creole5 models · code ht
- Hausa5 models · code ha
- Hawaiian5 models · code haw
- Kannada5 models · code kn
- Kazakh5 models · code kk
- Khmer5 models · code km
- Lao5 models · code lo
- Latin5 models · code la
- Lingala5 models · code ln
- Luxembourgish5 models · code lb
- Malagasy5 models · code mg
- Malayalam5 models · code ml
- Māori5 models · code mi
- Marathi5 models · code mr
- Norwegian Nynorsk5 models · code nn
- Occitan5 models · code oc
- Pashto5 models · code ps
- Punjabi5 models · code pa
- Sanskrit5 models · code sa
- Shona5 models · code sn
- Sindhi5 models · code sd
- Sinhala5 models · code si
- Somali5 models · code so
- Sundanese5 models · code su
- Tajik5 models · code tg
- Tamil5 models · code ta
- Tatar5 models · code tt
- Telugu5 models · code te
- Tibetan5 models · code bo
- Turkmen5 models · code tk
- Uzbek5 models · code uz
- Welsh5 models · code cy
- Yoruba5 models · code yo
- Afar1 model · code aa
- Irish1 model · code ga
- Kalaallisut1 model · code kl
- Norwegian Bokmål1 model · code nb
Glossary (24)
- ASR / speech-to-text (STT)automatic speech recognition
- Open weights
- Model card
- WERword error rate
- CERcharacter error rate
- Text normalization
- Hallucination
- Real-time factor (RTF)
- Latency
- VRAM and unified memory
- Quantization
- Native streaming
- Chunked (pseudo-)streaming
- VADvoice activity detection
- Speaker diarization
- Timestamps (segment and word level)
- Forced alignment
- Language identification
- Code-switching
- Punctuation and capitalization
- Fine-tuning and adaptation
- Encoder-decoder (attention) model
- CTCconnectionist temporal classification
- Transducer (RNN-T, TDT)
About this site (4)
Nothing matches. Try a model family (Whisper, Parakeet), a language, or a need such as “streaming” or “diarization”.