Ukudluliswa kwe-Free isiSwahili
Bhala umsindo isiSwahili kanye nevidiyo ibe ngumbhalo nge-AI. Ishesha, ilungile, futhi imahhala.
Indlela esebenza ngayo
- Iya ku- Umshicileli Free.ai
- Layisha phezulu ifayela lakho lomsindo noma levidiyo le-isiSwahili
- I-AI yethu ithola ngokuzenzakalela i-isiSwahili futhi ibhalela
- Layisha ngezansi incwadi yakho njengesihloko noma isizinda se-SRT
isiSwahili Izici zokudlulisa
- ✓Isebenza nge faster-whisper (ilayisense le-MIT)
- ✓Ukuqapha ulwimi oluzenzakalelayo isiSwahili
- ✓Isekela i-MP3, i-WAV, i-MP4, i-M4A, i-FLAC, nezinye
- ✓Ama-timestamps kanye ne-subtitle export (SRT)
- ✓Akukho kuphikiswa kwesayizi yefayela kuma-plans akhokhelwayo
- ✓Imfihlo nokuphepha -- amafayela asuswa ngemuva kokusebenza
Iminingwane yesilimi
| Ilimi | isiSwahili |
| Ikhodi ISO | sw |
| Imodeli AI | faster-whisper |
| Intengo | Ikhululekile |
Izilimi Eziningi
Bona zonke izilimiImibuzo ebuzwa kaningi
isiSwahili yimithombo encane yesilimi se Whisper - enkulu-v3-turbo ihlala ngaphezu kwe-25% yephutha legama, ngamanye amaxesha phezulu kakhulu. I-transcript isebenziseka ukukhangela ne-gist kodwa akufanele ithathwe njengeshicilelwa-kulungile. Uma i-engine ephakeme-yokunemba itholakala ku-isiSwahili siyixhuma ngokuzenzakalela.(Izinga D, over 25% word error rate kusethingi ze-benchmark - sishicilela izigaba ze-WER ezithembekile ngaphezu kokubhekwa kwe-marketing.)
Yebo - i-isiSwahili transcription iqala ngokudlulisa kusuka ku-token pool yakho yamahhala yansuku zonke. Umsindo ubiza ama-token angama-50 ngomzuzu, ngakho-ke i-token pool engaziwayo ihlanganisa amahora ambalwa we-audio ngosuku. Ama-akhawunti abhalisiwe athola u-token pool omkhulu wama-30,000 ngosuku. Ngaphakathi, ukudlulisa kukhokha-njengoku-u-ya-go, nge-token top-ups kusuka ku-$1.
isiSwahili izibhalo ezibhalwe phansi zibuyiselwa ku UTF-8 ejwayelekile nge-ortography ejwayelekile yesilimi.
MP3, WAV, M4A, FLAC, OGG, OPUS, ne WEBM zivunyelwe ngokuqondile. Ividiyo (MP4, MOV, MKV) siyisusa i-audio track server-side ngaphambi kokuyithumela ku-Whisper - awudingi ukushintsha noma yini ngokwakho. I-pipeline efanayo ngaphandle komthombo we-language, kufaka phakathi i-isiSwahili.
U-Anonymous ufaka i-upload engaphezu kuka-500 MB ngefayela ngalinye. Ama-akhawunti abhalisiwe afinyelela ku-2 GB. Usuku alukho umkhawulo onzima - amafayela ade ahlukaniswa ngokuzenzakalela (amafasitela amamitha ayi-30 ahlukaniswe) futhi aphinde ahlukaniswe ku-transcript eyodwa nesikhathi esiqhubekayo. Ukurekhodwa kwehora elilodwa isiSwahili (podcasts, izifundo ezigcwele, izinhlanganiso) kusebenza kahle.
Yebo - ukudweba umsindo kufakwe ngokuzenzakalela kuwo wonke ama-isiSwahili transcript. I-output ihlukaniswe njenge-Speech 1 / Speaker 2 / Speaker 3 ngesikhathi sosuku, ngakho-ke izingqungquthela, izingqungquthela zepaneli, kanye nezingqungquthela zeqembu eliningi zibuyela emuva zibhalwe. Ukudweba umsindo kuqhutshwa ngemodeli ehlukile futhi kusebenza ngokufanayo kuwo wonke ama-languages esiwaxhasayo.
Yebo - chofoza i-URL ku /transcribe/youtube/ ye-YouTube noma /transcribe/podcast/ ye-podcast feeds (Apple, Spotify, RSS). Silanda umsindo, siwuqhube nge-Whisper nge-language=sw, futhi sibuyisela umbhalo nge-timestamps ne-speaker labels. I-typical isiSwahili content: I-WhatsApp voice notes, i-YouTube vlogs, nevidiyo encane yizinto ezivame kakhulu isiSwahili workloads — chofoza i-URL ku /transcribe/youtube/ noma ulayishe umsindo ngqo.
Whisper ibiza cishe ama-token angama-50 ngomzuzu wesandi, ngakho-ke ukurekhodwa kwehora elilodwa kubiza ama-token angama-~3,000. Abaningi abasebenzisayo abachithanga lutho - i-pool yamahhala yansuku zonke yama-token angama-30,000 ifaka ama-clip aphansi, ama-voice notes, nama-podcasts afanayo. Ngaphesheya kwayo, ukudluliswa kwenziwa ngemali-njengoba-uhamba, nge-token top-ups kusuka ku-$1.
Yebo - zombili izigaba-level (zozo zonke ~10-30 imizuzwana) kanye negama-level timestamps zikhona. Igama-level yiphutha le VTT/SRT subtitle export ngakho ama-captions asynchronize line-by-line. On the API set timestamps="word" in the request body. isiSwahili izibhalo ezibhalwe phansi zibuyiselwa ku UTF-8 ejwayelekile nge-ortography ejwayelekile yesilimi.
Yebo. I-POST umsindo (ingxenye eminingi/ifomu-data, igama lendawo "ihele") ku /v1/transcribe / nge-language=sw - noma ushiye i-language parameter ukuze i-Whisper ikwazi ukukhomba ngokuzenzakalela. Ibuyisela i-JSON ne-transcript, ama-segments, ama-timestamps, nama-speaker labels. Ubufakazi obugcwele kanye ne-SDK snippets ku /api/.
Yebo - uma isingeniso siqediwe, chofoza Ukuhumusha noma ubeke umbhalo ku /isingeniso/. isiSwahili ixhumana namanye amagama asizoxhasa (200+). Usuku lokuxoxa lidlulisa isingeniso /ukusho/; ukudlulisa ukuthumela ku /uzwi/tts/ ukuveza umsindo kulesizinda solimi.
Whisper's noise training helps less at this tier - the bottleneck is the amount of isiSwahili audio Whisper saw during training, not noise. Clean studio audio still beats noise audio, but neither will reach the accuracy you would get on a high-source language.Uma i-transcript ibuyela ingekho, thumela i-imeyili contact@free.ai ngefayela - sizobuyisela ama-token futhi sibheke ukuthi ngabe i-engine ehlukile iphatha umsindo wakho kahle.