Read the result

GET/v1/jobs/{id}

Field	Type	Description
stems[].name	string	Duplex Conversation Separation: `_spk1`, `_spk2`. Multi Mic Separation: one per input. Target Speaker Extraction: `target`. Noise Reduction: `clean`.
stems[].si_sdr	float \| null	Estimate in dB. Null without `scoring`, and on Multi Mic Separation, Target Speaker Extraction and Noise Reduction. Superseded by `quality`.
quality.si_sdr_estimated_db	float \| null	The same estimate, with its source named.
quality.si_sdr_measured_db	float \| null	A reference measurement. Null on every customer job.
quality.si_sdr_source	string \| null	`blind_estimate` or `reference_measured`.
quality.resynthesis	string[]	Steps that re-synthesized the signal (Bandwidth extension).
sample_rate	int	Duplex Conversation Separation: 48000 or 16000 · Multi Mic Separation: 48000 or 16000 · Target Speaker Extraction: 16000 · Noise Reduction: the input's rate, 16000 at least. Read the field.
billed_seconds, billed_usd	float	What was charged. Success only.

bash

curl -H "Authorization: Bearer $DUETA_API_KEY" https://dueta.ai/v1/jobs/$JOB_ID |
  jq '.result'

Response

json

{
  "status": "succeeded",
  "stage": "done",
  "progress": 1,
  "billed_seconds": 612.4,
  "billed_usd": 5.1033,
  "paid_by": { "kind": "personal" },
  "result": {
    "stems": [
      { "name": "conversation-00_vocals_spk1_enh_48k", "si_sdr": 14.8 },
      { "name": "conversation-00_vocals_spk2_enh_48k", "si_sdr": 14.8 }
    ],
    "sample_rate": 48000,
    "quality": { "si_sdr_estimated_db": 14.8, "si_sdr_measured_db": null,
                 "si_sdr_source": "blind_estimate",
                 "resynthesis": ["super_resolution"] }
  }
}

// A Multi Mic Separation job: one stem per input, named after it.
{ "result": { "stems": [ { "name": "alice-00", "si_sdr": null },
                         { "name": "bob-01",   "si_sdr": null } ],
              "sample_rate": 48000 } }

Download a stem

GET/v1/jobs/{id}/stems/{name}?format=wav|flac

Download wav, flac or format=preview (AAC). No credit purchase required.

bash

# WAV, byte for byte as rendered:
curl -fL -H "Authorization: Bearer $DUETA_API_KEY" -o stem.wav \
  "https://dueta.ai/v1/jobs/$JOB_ID/stems/$STEM_NAME"

# FLAC, the same samples losslessly at about half the size:
curl -fL -H "Authorization: Bearer $DUETA_API_KEY" -o stem.flac \
  "https://dueta.ai/v1/jobs/$JOB_ID/stems/$STEM_NAME?format=flac"

Download everything

GET/v1/jobs/{id}/archive?format=wav|flac

format defaults to wav; flac zips the FLAC of every stem. Streamed, so there is no Content-Length. The download is named dueta-<first 8 of the job id>-stems.zip, or -stems-flac.zip for FLAC.

bash

# Streamed, so there is no Content-Length; progress is what has arrived.
curl -fL -H "Authorization: Bearer $DUETA_API_KEY" -o result.zip \
  "https://dueta.ai/v1/jobs/$JOB_ID/archive"

Download the original

GET/v1/jobs/{id}/inputs/{name}

name is inputs[].stem or the stored filename.

bash

# name is inputs[].stem ("conversation-00") or the stored filename.
curl -fL -H "Authorization: Bearer $DUETA_API_KEY" -o original.wav \
  "https://dueta.ai/v1/jobs/$JOB_ID/inputs/conversation-00"

Cases

404: Not succeeded, unknown stem name, or pruned.
422: format is not wav or flac. There is no MP3.
503 / 504: FLAC failed or timed out. WAV is unaffected.
Retention: Files kept 7 days after the job; after a purchase (demo and full refunds excluded), until you delete them: DELETE /v1/jobs/{id}.
Redirect: Follow redirects: a stem may 302 to object storage.

Read the result

GET/v1/jobs/{id}

Field	Type	Description
stems[].name	string	Duplex Conversation Separation: `_spk1`, `_spk2`. Multi Mic Separation: one per input. Target Speaker Extraction: `target`. Noise Reduction: `clean`.
stems[].si_sdr	float \| null	Estimate in dB. Null without `scoring`, and on Multi Mic Separation, Target Speaker Extraction and Noise Reduction. Superseded by `quality`.
quality.si_sdr_estimated_db	float \| null	The same estimate, with its source named.
quality.si_sdr_measured_db	float \| null	A reference measurement. Null on every customer job.
quality.si_sdr_source	string \| null	`blind_estimate` or `reference_measured`.
quality.resynthesis	string[]	Steps that re-synthesized the signal (Bandwidth extension).
sample_rate	int	Duplex Conversation Separation: 48000 or 16000 · Multi Mic Separation: 48000 or 16000 · Target Speaker Extraction: 16000 · Noise Reduction: the input's rate, 16000 at least. Read the field.
billed_seconds, billed_usd	float	What was charged. Success only.

bash

curl -H "Authorization: Bearer $DUETA_API_KEY" https://dueta.ai/v1/jobs/$JOB_ID |
  jq '.result'

Response

json

{
  "status": "succeeded",
  "stage": "done",
  "progress": 1,
  "billed_seconds": 612.4,
  "billed_usd": 5.1033,
  "paid_by": { "kind": "personal" },
  "result": {
    "stems": [
      { "name": "conversation-00_vocals_spk1_enh_48k", "si_sdr": 14.8 },
      { "name": "conversation-00_vocals_spk2_enh_48k", "si_sdr": 14.8 }
    ],
    "sample_rate": 48000,
    "quality": { "si_sdr_estimated_db": 14.8, "si_sdr_measured_db": null,
                 "si_sdr_source": "blind_estimate",
                 "resynthesis": ["super_resolution"] }
  }
}

// A Multi Mic Separation job: one stem per input, named after it.
{ "result": { "stems": [ { "name": "alice-00", "si_sdr": null },
                         { "name": "bob-01",   "si_sdr": null } ],
              "sample_rate": 48000 } }

Download a stem

GET/v1/jobs/{id}/stems/{name}?format=wav|flac

Download wav, flac or format=preview (AAC). No credit purchase required.

bash

# WAV, byte for byte as rendered:
curl -fL -H "Authorization: Bearer $DUETA_API_KEY" -o stem.wav \
  "https://dueta.ai/v1/jobs/$JOB_ID/stems/$STEM_NAME"

# FLAC, the same samples losslessly at about half the size:
curl -fL -H "Authorization: Bearer $DUETA_API_KEY" -o stem.flac \
  "https://dueta.ai/v1/jobs/$JOB_ID/stems/$STEM_NAME?format=flac"

Download everything

GET/v1/jobs/{id}/archive?format=wav|flac

bash

# Streamed, so there is no Content-Length; progress is what has arrived.
curl -fL -H "Authorization: Bearer $DUETA_API_KEY" -o result.zip \
  "https://dueta.ai/v1/jobs/$JOB_ID/archive"

Download the original

GET/v1/jobs/{id}/inputs/{name}

name is inputs[].stem or the stored filename.

bash

# name is inputs[].stem ("conversation-00") or the stored filename.
curl -fL -H "Authorization: Bearer $DUETA_API_KEY" -o original.wav \
  "https://dueta.ai/v1/jobs/$JOB_ID/inputs/conversation-00"

Cases

404: Not succeeded, unknown stem name, or pruned.
422: format is not wav or flac. There is no MP3.
503 / 504: FLAC failed or timed out. WAV is unaffected.
Retention: Files kept 7 days after the job; after a purchase (demo and full refunds excluded), until you delete them: DELETE /v1/jobs/{id}.
Redirect: Follow redirects: a stem may 302 to object storage.

Get results

Read the result

Download a stem

Download everything

Download the original

Cases

Get results

Read the result

Download a stem

Download everything

Download the original

Cases