Chidzidzo cheAI yeRuzha

EnCodec Audio Compression

EnCodec ndeye Meta's high-fidelity neural audio codec inodzvanya kutaura uye mimhanzi payakaderera bitrate ine mhando inokwikwidza inorema mafomati.

Teerera kugwaro iriAI-yakarondedzerwa · ~2 min

Pa peji ino2 min verenga
  1. Pfupiso
  2. Kudzika Kwakadzika
  3. Kuitwa muNyika yeChokwadi
  4. Ramangwana reEnCodec Audio Compression
  5. Kuzvishandisa
  6. Ramba Uchiongorora
  7. Mibvunzo inowanzo bvunzwa

Pfupiso

Izvo zvine basa nekuti inosimbisa yemazuva ano inobereka maodhiyo masisitimu uye ngarava mune yakavhurika-sosi fomu kuti chero munhu ashandise.

Kudzika Kwakadzika

Yakaburitswa ne Meta AI muna 2022, EnCodec inotevera SoundStream blueprint ye encoder, yasara vector quantizer (RVQ), uye decoder yakadzidziswa kupera, asi inowedzera zvakati wandei. Iyo inoshandisa kutenderera-inokwanisa convolutional encoder, yakawanda-scale spectrogram uye nguva-domain kuvakazve kurasikirwa, uye vanopikisa vanosarura kune yemhando yekunzwisisa. Mupiro unozivikanwa idiki Transformer-based entropy modhi iyo inowedzera kudzvanya macode akaverengerwa zvisina kurasikirwa, kudzvanya mabheti kunze pasina kurasikirwa kwemhando. EnCodec inosvitsawo chiyero chinoyera otomatiki akawanda anokwikwidza kurasikirwa kwekudzidziswa kuti vagare vakagadzikana. Inobata 24 kHz monophonic uye 48 kHz stereo odhiyo, inoshanda pamabitrates akadai se1.5, 3, 6, uye 12 kbps, uye pa6 kbps inosvika pamhando inofananidzwa neMP3 pa64 kbps. Zviratidzo zvayo zvine simba Meta's MusicGen uye AudioGen.

Kuitwa muNyika yeChokwadi

Tokening audio ye Meta's MusicGen uye AudioGen mavara-kune-audhiyo jenareta

Kudzvanya 24 kHz kutaura kune 1.5-6 kbps kune bandwidth-shoma kutapurirana

Encoding 48 kHz stereo mimhanzi ine mhando padyo neMP3 pakakwirira zvakanyanya bitrate

Kushanda seyakavhurika-sosi yekudonha-in codec yekutsvagisa uye odhiyo ML mapaipi kuburikidza neakaburitswa macheki

Ramangwana reEnCodec Audio Compression

EnCodec yatove iyo yekusarudzika tokenizer kune akati wandei akavhurika generative odhiyo modhi, uye vazukuru vayo vari kusundira kutendeka kwepamusoro pakaderera bitrate, yakazara stereo uye mimhanzi-giredhi kuvakazve, uye yakasimba kusanganisa nemavara-ku-odhiyo uye zvinyorwa-kumimhanzi majenareta. Tarisira kugamuchirwa kwakakura mukutaurirana kwakaderera-bandwidth, kutenderera-chaiyo-nguva, uye seyakajairwa 'audio token' dhizaini inoita kuti mavakirwo makuru emutauro-modhiyo averenge nekunyora ruzha.

Kuzvishandisa

Nei zvichikosha

  • Svika uye svika. Inonatsiridza kusvikika kuburikidza nekunyora, kurondedzera, uye mazwi ekubatanidza.
  • Mutengo uye bhajeti. Zvikwata zveMedia zvinogona kutumira odhiyo yakakwenenzverwa nekukurumidza nemabhajeti madiki.
  • Kumhanya uye chiyero. Masisitimu anotarisana nevatengi anogona kugadzirisa kutaurirana kwekutaura pamwero mukuru.

Chenjera

  • Kushandisa izwi zvisizvo uye njodzi dzekuedzesera dzinowedzera kana chibvumirano chisipo.
  • Kururama kunogona kudonha mumitauro, mataurirwo, kana nharaunda dzine ruzha.
  • Synthetic audio inogona kukanganisa kutaura kwechokwadi isina mavara akajeka.

Matanho anotevera

  1. Wana mvumo yakajeka yekutora inzwi, kugadzira, uye kushandisa zvakare.
  2. Yedza mhando pavatauri vakasiyana uye mamiriro ekumashure.
  3. Tsanangura apo munhu anofanira kuongorora kana kubvumidza zvabuda.
  4. Label synthetic odhiyo uye chengetedza marekodhi ekuzvidavirira.

Ramba Uchiongorora

Tsamba yemahara

Tora iyo yemazuva ese AI pfupiso

Nhatu dzakasimbiswa dzeAI nyaya mangwanani ega ega pakati pevhiki, dzakanyorwa neChirungu chakajeka. Yemahara zvachose, hapana ads.

Imwe email zuva rega rega revhiki. Bvisa kunyoresa kamwe chete. Hatimbotengese kana kugovera kero yako.

Zviedze iwe pachako

Take the EnCodec Audio Compression quiz

Mhinduro yepakarepo pamhinduro yega yega, uye chitupa chinogovaniswa chine ID inobatika kana wapedza kosi.

Tanga mibvunzo

Tsigira dzidzo yemahara yeAI. AI Understanding ndeye 501(c)(3) isingabatsiri - hapana ads, hapana paywall, zvachose. Ita chipo

Mibvunzo inowanzo bvunzwa

Chii chinonzi EnCodec Audio Compression?

EnCodec ndeye Meta's high-fidelity neural audio codec inodzvanya kutaura uye mimhanzi payakaderera bitrate ine mhando inokwikwidza inorema mafomati. Izvo zvine basa nekuti inosimbisa yemazuva ano inobereka maodhiyo masisitimu uye ngarava mune yakavhurika-sosi fomu kuti chero munhu ashandise.

Nderipi sangano rakaburitsa EnCodec?

EnCodec yakaunzwa ne Meta AI (FAIR) muna 2022 seyakanyanya kuvimbika neural audio codec.

Ndechipi chimwe chikamu chinowedzerwa neEnCodec pakudzvanya zvisingaite kubva kumatokeni ayo?

EnCodec inodzidzisa diki Transformer kufanotaura zvingangoitika uye arithmetic-makodhi, kuwana imwe yekusarasikirwa yekumanikidza pamusoro peRVQ.

Painenge 6 kbps, mhando yeEnCodec yakanzi inofananidzwa neMP3 pane iyo bitrate?

EnCodec pa 6 kbps inosvika kuzviisa pasi pemhando yakafanana neMP3 pa64 kbps, kuwana kukuru kwekubudirira.

Nderipi dambudziko rinogadziriswa neEnCodec's 'balancer' panguva yekudzidziswa?

Nekurasikirwa kwakawanda (kuvakazve, spectral, mhandu), iyo balancer inodzoreredza magradients avo saka hapana chinangwa chimwe chete chinotonga, kudzikamisa kudzidziswa.

Ndeipi nzira yakakosha yekuwedzera iyo EnCodec inogovana neSoundStream?

Kufanana neSoundStream, EnCodec inoshandisa yasara vector quantization kusiyanisa encoder embeddings mune akaturikidzana codebook indices.