Distil-Whisper
STT.ai haina kutamba iyi model. Iyi peji ndeyokuongorora ruzivo nezve iyo.
Transcribe ne models isu tinoita run →
5.8%
WER
99
Zvinhu
48.0x
_Kukurumidza
MIT
License
Chii Distil-Whisper
Distil-Whisper itsva yeWhisper yakagadzirwa neHugging Face. Iyo inoderedza saizi yemufananidzo ne49% uye inowana 6x yakanyanyisa kufungidzira, ichichengeta 1% WER yeiyo yekutanga Whisper Large V2 pane kunze kwezvirongwa zvekuongorora.
Zvinhu zvinotsigirwa ne Distil-Whisper
Model Info
- MugadziriHugging Face
- Architecture-
- LicenseMIT
- YakagadziridzwaMar 2026
Zvimwe zvinobvunzwa kakawanda
Distil-Whisper imhando yemutauro-ku-mutauro yakagadzirwa neHugging Face. STT.ai haina kutamba Distil-Whisper parizvino — iyi peji inoratidza ruzivo nezvemutauro. Kuti uite transliteration paSTT.ai, mutauro unotora unopa Whisper (Turbo, Large V3, Medium, Small).
Real-world kunyatsonzwisisa kunoenderana nemhando yezwi, accent, uye rurimi; For noisy kana accented rekodhi, kutarisira zvishoma zvidimbu zviviri zvemazana yepamusoro WER.
Distil-Whisper haisi kuwanikwa pa STT.ai. Yayo pachayo chibvumirano chekubvumidzwa zvinodzora kumisikidza iwe pachako. STT.ai's free tier inobata Whisper mamodheru isu tinoita kumisikidza — 600 maminitsi kuti awane, uyezve, pamwedzi, hapana kubhadharwa.
Distil-Whisper inoburitswa pasi peMIT, iyo yakavhurika-chigadzirwa chigadzirwa chitupa. Iwe unogona kuisa Distil-Whisper pane yako hardware kana kushandisa yedu inochengetwa vhezheni. Imwe neimwe inoshandiswa mukutengesa.
Distil-Whisper inotsigira 99 mazita ezvinyorwa. Kuwana otomatiki kunosarudza zvirinani zita rezvinyorwa zvemavhidhiyo; unogonawo kuisarudza nemunhu kuti uwane kunyatsoita kwakanaka.
Distil-Whisper inogadzirisa audio pazvinosvika 48.0x real-time pane edu GPUs. 1-hour audio file inosvika pasi pe 1 maminitsi; zvinopfuura zvinyorwa zvinomirira uye zvinozivisa neemail kana zvaitwa.
Distil-Whisper ine 756M parameters. Larger models tend to be more accurate but slower; STT.ai hosts Distil-Whisper on GPU so the parameter count doesn't affect your client-side performance.
Distil-Whisper inogamuchira ese mafomati STT.ai anotsigira - MP3, WAV, M4A, FLAC, OGG, MP4, MKV, MOV, WebM, AVI, uye vamwe.Kuburitsa se TXT, SRT, VTT, DOCX, JSON, kana PDF.
Yega. Mutauro diarization inofamba pamwe Distil-Whisper yese transcription — mumwe mutauro iri rakanyorwa uye unogona kuchinja zita ravo mu editor mushure.
Yeah. Distil-Whisper runs in our managed environment — audio is processed and deleted by default and never used for training without explicit opt-in. Pro plans add client-side encryption for transcripts at rest.
Usati washandisa Distil-Whisper, shandisawo Distil-Whisper vs Whisper Large V3 kuongorora kuti uone kuti Distil-Whisper inoenderana nemhando ipi neipi yefoni. Unogona kuona WER, segment count, speaker labels, uye confidence scores side-by-side. Distil-Whisper vs Whisper Large V3 kuongorora ndiyo inonyanya kushandiswa.
Yeah. Chidza "distil-whisper" separameter yemufananidzo pa /v1/transcribe endpoint. Python uye Node.js SDKs dzinosanganisira Distil-Whisper mifananidzo. Yemahara API tier inosanganisira 100 maminitsi / mwedzi.
Yeah. Nekudaro, nekuti Distil-Whisper ine MIT-lisensi, unogona kuichengeta iwe pachako. STT.ai's open-source peji rinonyora repo uye zviyero zveprojekiti. Zvikwata zvakawanda zvekugadzira zvinoshandisa yedu inochengetwa vhezheni kuti urege kubhadharisa GPU, kuchinja mamodheru, uye ops.