Whakapāpāhotanga Huitene wātea
Ka whakamāoritia te oro me te ataata Huitene ki te kupu me AI. Āhua tere, tika, me te wātea.
He pēhea te mahi
- Tere ki te Free.ai Transscriber
- Whakarewa i tōmu pūoro Huitene, he pūrongo pūoro rānei
- Ka kitea ā-māoritia e tātau AI te Huitene, ā, ka whakamāoritia
- Whakahua i tōna whakahua hei kupu, hei whakahua SRT rānei
Huitene He āhuatanga whakamāoritanga
- ✓Ka whakahaua e faster-whisper (MIT licensed)
- ✓Kitenga reo Huitene ā-mātau
- ✓E tautoko ana i te MP3, WAV, MP4, M4A, FLAC, me ētahi atu
- ✓Ko ngā tātai wā me te whakarerekētanga o te whakarārangi (SRT)
- ✓Kāore he tepe o te rahi o te pūkete i runga i ngā mahere utu
- ✓Private me te haumaru -- ka whakakoretia ngā faila i muri i te tukanga
Whakamāramatanga o te reo
| reo | Huitene |
| Kāri ISO | sv |
| Tauira AI | faster-whisper |
| Whakahauhau | Waihoki |
He maha atu reo
Tirohia ngā reo katoaFAQ
Whisper te nui-v3-turbo e whakahaere ana i te Huitene - 7-15% te mokatere hapa kupu i runga i te reo tohu. E tūmanako ana ki ngā whakakapi i ngā mea whakaingoatanga, ngā tau, me ngā kupu hangarau mārō; ka tika te nuinga o te whakahua.(Tier B, 7-15% word error rate i runga i ngā huinga tohu - ka whakaputaina e tātau ngā taumata pono WER kaua ko ngā take mākete.)
He - ko te whakamārōtanga Huitene e whakataki ana i te tuatahi mai i tō tātou pūpū tāuru wātea i ia rā. Ko te pūoro e utu ana i ngā tāuru 50 i ia minu, nō reira ko te pūoro ā-rohe i ia rā e taupoki ana i ētahi wā o te oro i ia rā. Ka whiwhi te pūrongo i te pūoro nui ake o te 30,000-tāuru i ia rā. I muri iho, ko te whakamārōtanga he utu-i-te-rohe, me ngā tāuru-i-te-rohe mai i te $1.
Ka hoki ngā whakarerekētanga Huitene ki te UTF-8 paerewa me te whakamāoritanga noa o te reo.
MP3, WAV, M4A, FLAC, OGG, OPUS, me WEBM e whakaaetia ana. Mō ngā ataata (MP4, MOV, MKV) ka tangohia e tātau te aho-pāpāho i mua i te tukuna ki Whisper - kāore e hiahiatia kia tahuri e koe. He ōrite te pūtautau ahakoa te reo pūtake, tae atu ki te Huitene.
Ko te whakaata whakaata ā-whakahaere i te āhua o te 500 MB ia pūkete. Ko ngā pūkete i whakaingoatia e tae atu ki te 2 GB. Kāore he tepe uaua te roanga - ka whakawāteatia ngā pūkete roa i te wā kotahi (30-sekondi matapihi me te whakawāteatanga) me te whakatata ki roto i tētahi pūkete kotahi me ngā wā whakatata. He pai te mahi o ngā pūkete Huitene maha-ora (pūkete, whakaakoranga katoa, hui).
He - ko te kaikōrero diarization i runga i te pūmau mō ia whakahuatanga Huitene. Ka whakawāteatia te huaputa hei Kaikōrero 1 / Kaikōrero 2 / Kaikōrero 3 me ngā tātaitai wā, nā reira ko ngā uiuinga, ngā kōrero papa, me ngā hui maha-pāpāho e hoki ana ki te tohu. Ka haere te Diarization i runga i tētahi tauira motuhake, ā, he ōrite te mahi i ngā reo katoa e tautoko ana e tātau.
He - whakataki i te URL ki roto i /transcribe/youtube/ mō YouTube, /transcribe/podcast/ rānei mō ngā pūngao podcast (Apple, Spotify, RSS). Ka whakataki tātau i te oro, ka haere i roto i te Whisper me te reo=sv, ā, ka hoki te whakarerekētanga me ngā tātai wā me ngā tohu kaikōrero. Ko ngā ihirangi Huitene noa iho: ngā whakaakoranga, ngā uiui, ngā tuhipoka reo, me ngā ihirangi YouTube i roto i te Huitene katoa e mahi ana - whakataki i tētahi URL ki roto i /transcribe/youtube/, ka whakatakatia rānei te faila i te āhua tonu.
Ko te utu o te Whisper tata ki te 50 ngā tohu i ia minu o te oro, nō reira ko te whakataki kotahi wā ~3,000 ngā tohu. Ko te nuinga o ngā kaiwhakamaori kāore i te pau i tētahi mea - ko te pūtea wātea o te 30,000 ngā tohu i ia rā e taupoki ana i ngā tākaro poto, ngā tuhipānui reo, me ngā podcast kotahi. I tua atu i tēnei, ko te whakamārōro he utu-i-te-rohe, me ngā tohu whakahauhau mai i te $1.
He - e rua ngā taumata wā o te wāhanga (i ia ~10-30 sekona) me ngā wā-wā e wātea ana. Ko te taumata-wā te pūmau mō te whakawāteatanga o te whakawāteatanga VTT/SRT kia ōrite ai ngā whakamāramatanga ki te rārangi-ki-rārangi. I runga i te API e whakaritea ana ngā wāhanga ="wā" i te tinana tono. Ka hoki ngā whakarerekētanga Huitene ki te UTF-8 paerewa me te whakamāoritanga noa o te reo.
He. POST audio (maha-wāhanga/whakahaere-whakahaere, ingoa āpure "file") ki te /v1/transcribe/ me te reo=sv - ka waiho rānei te tohuāhua reo kia taea ai te Whisper te kite i te āhua o te āhua. Ka hoki mai te JSON me te whakamāoritanga, ngā wāhanga, ngā tātaitai wā, me ngā tohu whakarongo. Ko te tohutoro katoa me ngā tātai SDK i te /api/.
He - i te mutunga o te whakamāoritanga, ka whakataki te whakamāoritanga, te whakarārangi rānei i te kupu ki / whakamāori /. Huitene ngā takirua me ētahi atu reo e tautoko ana i a mātau (200+). Mō ngā wā o te hui, ka whakawāteatia te whakamāoritanga mā / whakarārangi /; mō te whakamātautau, ka tukuna ki / oro/tts/ hei whakapāpāho i te oro i te reo ūnga.
Kua whakaakona te Whisper i runga i ngā rau mano o ngā wā o te oro o te ao tūturu, nā reira ka whakaaetia e ia te mārō o te papamuri me ngā pūkete mō te huapu i runga i te Huitene. Mō ngā hua pai rawa, ka whakarato āhua oro mārō (hearing microphone, kāore he pūoro) - i tēnei taumata ka whakakotahi te mārō ki te mokatere hapa taketake. If a transcript comes back unusable, email contact@free.ai with the file - we will refund the tokens and look at whether a different engine handles your audio better.