In your cmd start the server with:
python -m speechcraft.server
Note: The first time you start the server, it will download the models. This can take a while.
The Webservice is built with FastTaskAPI. In this regard, for each request it will create a task and return a job id. You can then check the status of the job and retrieve the result.
We recommend using fastSDK and FastTaskAPI as illustrated for easy file transfers.
Read the documentations of fastSDK, FastTaskAPI and media-toolkit to get the most out of the service and to familiarize yourself with the concepts.
You can configure some settings via environment variables:
| ENV Variable | Description |
|---|---|
| MODELS_DIR | Path to the folder where the models are stored. For example the inswapper and the GPEN models are downloaded and stored in this folder. |
| EMBEDDINGS_DIR | Path to the folder where the face embeddings are stored. Stored faces can be reused by the api. |
| ALLOW_EMBEDDING_SAVE_ON_SERVER | If set to True, the embeddings can be saved on the server. Consider to set it false to avoid memory overflow or in multi-user scenarios |
For more settings and how to deploy the service check-out FastTaskAPI. For example it allows you to deploy the service with Runpod out of the box.
The webservice provides enpdoints for the swap, add_face and enhance_face functionality. You can send requests to the endpoints with any http client, e.g. requests, httpx, curl, etc.
Using fastSDK is the most convenient way to interact with the webservice. You can find an implementation of an SDK generated for fastSDK in the socaity SDK documentation.
FastSDK is a python module that provides a convenient way to interact with the webservice. You can find an implementation of an SDK generated for fastSDK in the socaity SDK documentation.
from media_toolkit import AudioFile
# text2speech
response = httpx.post("http://localhost:8009/text2voice", params={ "text" : "please contribute", "voice": "en_speaker_3"})
# voice cloning
audio = AudioFile().from_file("myfile.wav")
request = httpx.post(
"http://localhost:8009/voice2embedding", params={ "voice_name" : "hermine"},
files={"audio": audio.to_httpx_send_able_tuple()}
)
# voice2voice
response = httpx.post(
"http://localhost:8009/voice2voice",
params={ "voice_name" : "hermine"},
files={"audio": audio.to_httpx_send_able_tuple()}
)This example shows how to do it with plain python requests.
import requests
# text2speech synthesis
response = requests.post("http://localhost:8009/text2voice", params={ "text" : "please contribute", "voice": "en_speaker_3"})
# Speaker embedding creation
with open("myfile.wav", "rb") as f:
audio = f.read()
response = requests.post(
"http://localhost:8009/voice2embedding",
params={ "voice_name" : "my_new_speaker"},
files={"audio_file": audio}
)
# voice2voice
response = requests.post(
"http://localhost:8009/voice2voice",
params={ "voice_name" : "my_new_speaker"},
files={"audio_file": audio}
)The response is a json that includes the job id and meta information. By sending then a request to the job endpoint you can check the status and progress of the job. If the job is finished, you will get the result, including the swapped image.
import requests
from media_toolkit import AudioFile
# check status of job
response = requests.get(f"http://localhost:8009/api/job/{job.json()['job_id']}")
# convert result to image file
audio = AudioFile().from_bytes(response.json()['result']))