Stable Audio 3 Trainer

Unganisha

Kitambulisho cha modeli stable-audio-3-trainer

# Run the model; wait up to 20 s for the finished result in the same call
curl -X POST "https://api.doitong.ru/v1/run/stable-audio-3-trainer?wait=20" \
  -H "Authorization: Bearer $DOITONG_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "audio_data_url": "https://content.doitong.com/audio/showcase/v3/tools/dub_es.mp3"
  }'

# Still "queued" or "running"? Poll with the id from the response:
curl "https://api.doitong.ru/v1/requests/REQUEST_ID?wait=20" \
  -H "Authorization: Bearer $DOITONG_API_KEY"

const API = "https://api.doitong.ru/v1";
const headers = {
  "Authorization": "Bearer " + process.env.DOITONG_API_KEY,
  "Content-Type": "application/json",
};

// Run the model; wait up to 20 s for the finished result in the same call
let res = await fetch(API + "/run/stable-audio-3-trainer?wait=20", {
  method: "POST",
  headers,
  body: JSON.stringify({
    "audio_data_url": "https://content.doitong.com/audio/showcase/v3/tools/dub_es.mp3"
  }),
});
let request = await res.json();
if (!res.ok) throw new Error(request.error.code + ": " + request.error.message);

// Not finished yet: long-poll until it is
while (request.status === "queued" || request.status === "running") {
  res = await fetch(API + "/requests/" + request.id + "?wait=20", { headers });
  request = await res.json();
}
if (request.status !== "succeeded") throw new Error(JSON.stringify(request.error));
console.log(request.outputs.map((o) => o.url), request.cost.credits + " API credits ($" + request.cost.usd + ")");

import os
import requests

API = "https://api.doitong.ru/v1"
headers = {"Authorization": "Bearer " + os.environ["DOITONG_API_KEY"]}

payload = {
    "audio_data_url": "https://content.doitong.com/audio/showcase/v3/tools/dub_es.mp3",
}

# Run the model; wait up to 20 s for the finished result in the same call
r = requests.post(API + "/run/stable-audio-3-trainer", params={"wait": 20}, json=payload, headers=headers)
request = r.json()
if not r.ok:
    raise RuntimeError(request["error"]["code"] + ": " + request["error"]["message"])

# Not finished yet: long-poll until it is
while request["status"] in ("queued", "running"):
    request = requests.get(API + "/requests/" + request["id"], params={"wait": 20}, headers=headers).json()
if request["status"] != "succeeded":
    raise RuntimeError(request["error"])
print([o["url"] for o in request["outputs"]], request["cost"]["credits"], "API credits", "$" + request["cost"]["usd"])

# 1. Connect once (Claude Code; Cursor, VS Code, Claude Desktop: https://doitong.com/sw/developers/docs/mcp)
claude mcp add --transport http doitong-models https://api.doitong.ru/v1/mcp \
  --header "Authorization: Bearer $DOITONG_API_KEY"

# 2. Your agent calls the "run" tool with:
{
  "model_id": "stable-audio-3-trainer",
  "input": {
    "audio_data_url": "https://content.doitong.com/audio/showcase/v3/tools/dub_es.mp3"
  }
}

Piga modeli ya Stable Audio 3 Trainer (stable-audio-3-trainer) kupitia Doitong API. Nyaraka za mawakala: https://doitong.com/sw/developers/models/stable-audio-3-trainer.md

POST https://api.doitong.ru/v1/run/stable-audio-3-trainer?wait=20
Authorization: Bearer $DOITONG_API_KEY
Content-Type: application/json

{
  "audio_data_url": "https://content.doitong.com/audio/showcase/v3/tools/dub_es.mp3"
}

Ikiwa hali ya jibu ni "queued" au "running", piga GET https://api.doitong.ru/v1/requests/REQUEST_ID?wait=20 mara kwa mara hadi iwe "succeeded"; faili ziko kwenye outputs[].url. POST https://api.doitong.ru/v1/quote/stable-audio-3-trainer yenye mwili uleule hurejesha bei kamili kabla ya kuendesha.

Faili kama mfano: $ 3.00 salio za API 3,000 · ≈ ₽ 316.39 faili zako hupimwa kabla ya kuendesha Iendeshe kwenye ukurasa huu

Pata ufunguo wa API

Stability AI Mafunzo 1–380 s #music#audio#sfx#lora

Hufunza LoRA kwenye Stable Audio 3 kutoka kwenye zip ya klipu za sauti zenye manukuu ya maandishi yanayolingana

OpenAPI Markdown

Stability AI · Faili

Tokeo lako la kwanza linaonekana hapa

Unda ombi hapa chini na uliendeshe kwa ufunguo wako wa API. Matokeo yanakiliwa kwenye hifadhi yetu, yakihifadhiwa milele au 7 siku bila malipo.

Unda ombi

Bei kwa kila uendeshaji

Bei kulingana na uingizaji wako

$ 3.00kwa utekelezaji

salio za API 3,000 · ≈ ₽ 316.39

Bei: kuanzia $ 0.003

utekelezaji 1 × $ 3.00 Endesha · $ 3.00

Hutozwa tu ikifanikiwa, na kurejeshewa kiotomatiki ikishindwa.

number_of_steps chaguomsingi 1000

batch_size chaguomsingi 1

Faili zako hupimwa kabla ya kuendeshwa (urefu, mwonekano, kasi ya fremu); makadirio hurejesha bei kamili na uendeshaji hutozwa kiasi hicho hicho.

Endpoint ya makadirio: POST /quote/{id}

Unda ombi

Vidhibiti vinazalishwa kutoka kwenye schema ya modeli. Msimbo na bei husasishwa unapovibadilisha.

audio_data_url •inahitajika

URL ya zip ya dataset, ikiwa kila jozi ya sauti/manukuu imehifadhiwa pamoja kwa majina ya msingi yanayolingana: `clip.wav` huambatana na `clip.txt`, kwa mfano. Manukuu lazima yatumie kiendelezi `.txt`.

number_of_steps chaguomsingi 1000

Idadi ya hatua za mafunzo ya LoRA.

duration

Muda wa crop/pad kwa sekunde, unaowekewa kikomo na urefu asilia wa mafunzo wa modeli iliyochaguliwa. Toa urefu au uache ili kutumia klipu ndefu zaidi iliyogunduliwa kwenye dataset, kwa kuzingatia kikomo hicho.

batch_size chaguomsingi 1

Ukubwa wa kundi unaotumika wakati wa mafunzo.

Ya kina model · base_precision · adapter_type · lora_checkpoint_url · learning_rate · rank · seed · pre_encode · include · exclude

model chaguomsingi medium-base

Checkpoint msingi wa kufanya fine-tuning ya Stable Audio 3.

base_precision chaguomsingi bf16

Usahihi unaotumiwa na uzani msingi uliogandishwa; vigezo vya LoRA hubaki fp32.

adapter_type chaguomsingi dora-rows

Familia ya adapta inayotumika kwa mafunzo ya LoRA.

lora_checkpoint_url

URL ya hiari ya checkpoint ya LoRA ya `.safetensors` ya kuendelezea kutoka hapo.

learning_rate chaguomsingi 0.0001

Kiwango cha kujifunza kinachotumiwa na AdamW kwenye vigezo vya LoRA.

rank chaguomsingi 16

Rank ya adapta ya LoRA.

seed chaguomsingi 42

Seed ya kubadilisha nasibu.

include

Sehemu za majina ya moduli zinazoweka kikomo mahali LoRA inaongezwa: moduli zinazolingana pekee ndizo zinazoipokea.

exclude

Sehemu za majina zinazotumika kuwatenga moduli zinazolingana.

POST /run/stable-audio-3-trainer?wait=20 makadirio → $ 3.00

# Run the model; wait up to 20 s for the finished result in the same call
curl -X POST "https://api.doitong.ru/v1/run/stable-audio-3-trainer?wait=20" \
  -H "Authorization: Bearer $DOITONG_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "audio_data_url": "https://content.doitong.com/audio/showcase/v3/tools/dub_es.mp3"
  }'

# Still "queued" or "running"? Poll with the id from the response:
curl "https://api.doitong.ru/v1/requests/REQUEST_ID?wait=20" \
  -H "Authorization: Bearer $DOITONG_API_KEY"

const API = "https://api.doitong.ru/v1";
const headers = {
  "Authorization": "Bearer " + process.env.DOITONG_API_KEY,
  "Content-Type": "application/json",
};

// Run the model; wait up to 20 s for the finished result in the same call
let res = await fetch(API + "/run/stable-audio-3-trainer?wait=20", {
  method: "POST",
  headers,
  body: JSON.stringify({
    "audio_data_url": "https://content.doitong.com/audio/showcase/v3/tools/dub_es.mp3"
  }),
});
let request = await res.json();
if (!res.ok) throw new Error(request.error.code + ": " + request.error.message);

// Not finished yet: long-poll until it is
while (request.status === "queued" || request.status === "running") {
  res = await fetch(API + "/requests/" + request.id + "?wait=20", { headers });
  request = await res.json();
}
if (request.status !== "succeeded") throw new Error(JSON.stringify(request.error));
console.log(request.outputs.map((o) => o.url), request.cost.credits + " API credits ($" + request.cost.usd + ")");

import os
import requests

API = "https://api.doitong.ru/v1"
headers = {"Authorization": "Bearer " + os.environ["DOITONG_API_KEY"]}

payload = {
    "audio_data_url": "https://content.doitong.com/audio/showcase/v3/tools/dub_es.mp3",
}

# Run the model; wait up to 20 s for the finished result in the same call
r = requests.post(API + "/run/stable-audio-3-trainer", params={"wait": 20}, json=payload, headers=headers)
request = r.json()
if not r.ok:
    raise RuntimeError(request["error"]["code"] + ": " + request["error"]["message"])

# Not finished yet: long-poll until it is
while request["status"] in ("queued", "running"):
    request = requests.get(API + "/requests/" + request["id"], params={"wait": 20}, headers=headers).json()
if request["status"] != "succeeded":
    raise RuntimeError(request["error"])
print([o["url"] for o in request["outputs"]], request["cost"]["credits"], "API credits", "$" + request["cost"]["usd"])

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "run",
    "arguments": {
      "model_id": "stable-audio-3-trainer",
      "input": {
        "audio_data_url": "https://content.doitong.com/audio/showcase/v3/tools/dub_es.mp3"
      },
      "wait_seconds": 20
    }
  }
}

Endesha ombi hili

Ufunguo wa API

Key yako inabaki kwenye kivinjari hiki: inatumwa kwa Doitong API pekee na kuhifadhiwa (localStorage) tu ukiweka tiki kwenye Kumbuka. Kumbuka katika kivinjari hiki Pata ufunguo wa API

Kuhusu Stable Audio 3 Trainer

Huboresha checkpoint ya msingi (medium-base, small-music-base au small-sfx-base) kwa kutumia zip ya `audio_data_url`, ambapo kila faili ya sauti ina manukuu ya `.txt` yenye jina linalofanana. Unaweza kurekebisha `number_of_steps` (hadi 20000), `learning_rate`, `rank` (1 hadi 256), `adapter_type` na `batch_size`, au kuendelea kutoka checkpoint ya `.safetensors`. Hurejesha uzani wa LoRA na faili ya usanidi.

Vigezo

JinaAina ChaguomsingiInaruhusiwa Maelezo

audio_data_urlinahitajika

Ainastring

modelsi lazima

Ainaenum

Chaguomsingimedium-base

medium-basesmall-music-basesmall-sfx-base

Checkpoint msingi wa kufanya fine-tuning ya Stable Audio 3.

number_of_stepssi lazima

Ainainteger

Chaguomsingi1000

1 – 20,000

Idadi ya hatua za mafunzo ya LoRA.

learning_ratesi lazima

Ainanumber

Chaguomsingi0.0001

≤ 0.01> 0

Kiwango cha kujifunza kinachotumiwa na AdamW kwenye vigezo vya LoRA.

ranksi lazima

Ainainteger

Chaguomsingi16

1 – 256

Rank ya adapta ya LoRA.

adapter_typesi lazima

Ainaenum

Chaguomsingidora-rows

loradoradora-rowsdora-colsboralora-xsdora-rows-xsdora-cols-xsbora-xs

Familia ya adapta inayotumika kwa mafunzo ya LoRA.

durationsi lazima

Ainanumber | null

1 – 380

batch_sizesi lazima

Ainainteger

Chaguomsingi1

1 – 8

Ukubwa wa kundi unaotumika wakati wa mafunzo.

seedsi lazima

Ainainteger

Chaguomsingi42

0 – 2,147,483,647

Seed ya kubadilisha nasibu.

base_precisionsi lazima

Ainaenum

Chaguomsingibf16

bf16bfloat16fp16float16

Usahihi unaotumiwa na uzani msingi uliogandishwa; vigezo vya LoRA hubaki fp32.

includesi lazima

Ainastring[]

Sehemu za majina ya moduli zinazoweka kikomo mahali LoRA inaongezwa: moduli zinazolingana pekee ndizo zinazoipokea.

excludesi lazima

Ainastring[]

Sehemu za majina zinazotumika kuwatenga moduli zinazolingana.

lora_checkpoint_urlsi lazima

Ainastring | null

URL ya hiari ya checkpoint ya LoRA ya `.safetensors` ya kuendelezea kutoka hapo.

pre_encodesi lazima

Ainaboolean

Chaguomsingifalse

Usimbaji wa awali wa hifadhi ya sauti kuwa latenti za SAME kabla ya mafunzo ya LoRA.

inahitajika

Tokeo

GET /requests/req_8k2m4q7x1z9c3v5b6n0p2r4t

{
  "id": "req_8k2m4q7x1z9c3v5b6n0p2r4t",
  "object": "request",
  "model": "stable-audio-3-trainer",
  "status": "succeeded",
  "created_at": "2026-10-20T12:00:00Z",
  "started_at": "2026-10-20T12:00:02Z",
  "completed_at": "2026-10-20T12:00:09Z",
  "queue_position": null,
  "cost": {
    "credits": 3000,
    "usd": "3.00",
    "charged": 3000,
    "refunded": 0
  },
  "input": {
    "audio_data_url": "https://content.doitong.com/audio/showcase/v3/tools/dub_es.mp3"
  },
  "output": {
    "lora_file": {
      "url": "https://content.doitong.com/dev-api/2026/10/20/req_8k2m4q7x1z9c3v5b6n0p2r4t/0-output.jpeg",
      "content_type": "image/jpeg",
      "file_name": "0-output.jpeg",
      "file_size": 1843200
    },
    "config_file": {
      "url": "https://content.doitong.com/dev-api/2026/10/20/req_8k2m4q7x1z9c3v5b6n0p2r4t/0-output.jpeg",
      "content_type": "image/jpeg",
      "file_name": "0-output.jpeg",
      "file_size": 1843200
    }
  },
  "outputs": [
    {
      "type": "file",
      "url": "https://content.doitong.com/dev-api/2026/10/20/req_8k2m4q7x1z9c3v5b6n0p2r4t/0-output.jpeg",
      "content_type": "image/jpeg",
      "file_size": 1843200,
      "stored": true
    }
  ],
  "error": null,
  "metrics": {
    "inference_seconds": 6.8
  },
  "urls": {
    "self": "https://api.doitong.ru/v1/requests/req_8k2m4q7x1z9c3v5b6n0p2r4t",
    "cancel": "https://api.doitong.ru/v1/requests/req_8k2m4q7x1z9c3v5b6n0p2r4t/cancel"
  }
}

Faili zako, chaguo lakoKila outputs[].url ni nakala kwenye hifadhi yetu: huhifadhiwa milele (hutozwa kwa GB kwa mwezi) au 7 siku bila malipo, kulingana na mpangilio wa akaunti yako au ?storage=.

Muundo uleule kwenye kila modelitype, url, content_type, width, height na duration, kwa picha, video na sauti. output ni kitu cha matokeo cha modeli yenyewe.

Sehemu za tokeo

lora_file: object {url, content_type, file_name, file_size}Uzani wa LoRA uliofunzwa kwa Stable Audio 3.
config_file: object {url, content_type, file_name, file_size}Metadata ya JSON inayoeleza uendeshaji wa mafunzo na modeli inayooana kwa inference.

Kwa mawakala wa AI: ukurasa huu katika Markdown unapatikana kwenye https://doitong.com/sw/developers/models/stable-audio-3-trainer.md. Modeli na miongozo yote: https://doitong.com/developers/llms.txt

Tokeo lako la kwanza linaonekana hapa

Unda ombi

Endesha ombi hili

Kuhusu Stable Audio 3 Trainer

Vigezo

Tokeo

Miundo zaidi: Mafunzo

ERNIE-Image Trainer

FLUX 2 [klein] 9b Base Trainer Tofauti: 2

Flux 2 Klein 9B Base Trainer Tofauti: 2

FLUX 2 Trainer Tofauti: 2