ConductSpeech
For research labs and universities

Run a whole study in one batch. Export a results table in one click.

ConductSpeech Research is a batch-first language sample analysis workflow built for studies. Submit many samples at once, merge study details, export a results table, and open the results in the stats tools you already use.

$49 per month · cancel any time · batch API · CSV / SPSS / SAS / notebook exports

What the Research tier gives you

Every Clinician feature, plus the batch, study detail, and export tools a study actually needs.

Batch the whole study, not one sample at a time

Submit up to 1,000 samples in a single API call. Get back a per-sample table plus means across the batch and for each group, ready to use in your analysis.

Exports your stats software already speaks

Drop straight into SPSS (.sps syntax + CSV), SAS (.sas syntax + CSV), a flat results table, or a starter Python notebook. No proprietary binary, no license-locked viewer.

API access with your own key

Generate revocable API keys in your account settings. Keys are SHA-256 hashed at rest and shown to you exactly once. Each key shows when it was last used.

Reproducibility you can publish

Every measure we compute (MLU, TTR, NDW, PGU, DSS-lite, IPSyn-lite, story word counts) and its rules are described on the Methodology page, so your reviewers can see how each number is counted.

Submit a study in one request

No SDK to install. Just curl, Python requests, R httr — whatever you already use.

curl -X POST https://engine.conductspeech.com/batch/analyze \
  -H "X-API-Key: csp_..." \
  -H "Content-Type: application/json" \
  -d '{
    "items": [
      { "itemId": "P01-pre",  "label": "P01 baseline", "ageMonths": 48,
        "utterances": [...] },
      { "itemId": "P01-post", "label": "P01 12wk",     "ageMonths": 51,
        "utterances": [...] },
      { "itemId": "P02-pre",  "label": "P02 baseline", "ageMonths": 50,
        "utterances": [...] }
    ]
  }'

Response: per-item metrics, per-item advanced scores, and cohort summary stats. See the full API reference for the schema.

Your data, in your stats package

Choose the export format and we'll generate the results table, import scripts, or notebook for you.

SPSS

Generates a .sps syntax file plus a flat CSV. Run the syntax in SPSS to import the CSV with the right variable types and labels — no manual data dictionary work.

SAS

Generates a .sas program that imports the CSV companion file. Drop the pair into your project directory and submit.

CSV

Plain results table with one row per sample, study detail columns, and every exported metric. Loads cleanly in R, Python, Stata, Excel, or anything that opens a CSV.

Notebook

Generates a starter Python notebook that loads the results table, summarizes core metrics, and creates cohort summaries when a group column is present.

Reproducibility, not a black box

We document every metric we compute and every formula we use. Reviewers and replicators get the same answers from the same input — that's the whole point.

MLU (morphemes & words)
Total morphemes ÷ utterances; Brown's Stages convention.
TTR / NDW
Lexical diversity, with different words counted by root (dog and dogs are one word).
PGU
Percent of counted utterances with no error code; only the clinician's SALT codes count.
DSS-lite
An approximation of the Developmental Sentence Score (Lee, 1974).
IPSyn-lite
30 forms scored out of 60, an approximation of Scarborough (1990).
Story words
Counts of words from the Narrative, Expository or Persuasive word lists, not a rubric score.

Researcher questions, answered

What does a batch request look like?

Submit a batch request with each sample's label, participant age, optional study details, and utterances. ConductSpeech returns per-sample metrics and a batch summary.

Where does my data live?

Recordings are deleted after transcription. A saved sample keeps the transcript and measures in your account, so you can export them again later. We don't sell, share, or use your data to train external models. We do use transcripts on our own HIPAA-ready servers to improve ConductSpeech, and you can turn that off in Data Sharing.

Can I use ConductSpeech for an IRB-approved study?

ConductSpeech can be used with de-identified language samples in research settings. We can provide a data flow description for your IRB application. We are not currently able to sign a BAA — talk to us if your protocol involves PHI.

How is the research tier different from Clinician?

Research adds batch analysis, study-detail-aware exports, SPSS/SAS import files, Python notebook export, and self-serve API keys. If you're running studies, those tools reduce the reformatting work between sample analysis and statistics.

Can I cite ConductSpeech in a paper?

Yes. Contact us at info@conductscience.com and we will provide a methods citation including the version of the analyzer used and links to the formulas.

Stop reformatting CSVs

$49/month. API access. Results table, SPSS, SAS, CSV, and notebook exports. Cancel any time.