Characterising DNA in specimens from people with tuberculosis (TB), a major cause of death, is critical for evaluating diagnostics and the microbiome, yet extracellular DNA, more frequent in people on chemotherapy, confounds results. We evaluated whether nucleic acid dyes [propidium monoazide (PMA), PEMAX] and DNaseI could reduce this. PCR [16S Mycobacterium tuberculosis complex (Mtb) qPCR, Xpert MTB/RIF] was done on dilution series of untreated and treated (PMA, PEMAX, DNaseI) Mtb. Separately, 16S rRNA gene qPCR and sequencing were done on untreated and treated patients' sputa before (Cohort A: 11 TB-negatives, 9 TB-positives; Cohort B: 19 TB-positives, PEMAX only) and after 24 weeks of chemotherapy (only Cohort B). PMA and PEMAX reduced PCR-detected Mtb DNA for both the dilution series and Cohort A sputum versus untreated controls, suggesting non-intact Mtb is present before treatment start. PEMAX enabled sequencing Mycobacterium-detection in 5/9 (59%; Cohort A) and 2/3 (67%; Cohort B week 0) TB-positive sputa where no reads otherwise occurred. In Cohort A, PMA- and PEMAX-treated versus untreated sputa had decreased α- and increased β-diversities. In Cohort B, α-diversity differences between untreated and PEMAX-treated sputa occurred only at 24-weeks and β-diversity differences between timepoints were only detected after PEMAX. DNaseI had negligible effects. PMA and PEMAX (but not DNaseI) reduced extracellular DNA, improving the proportion of Mycobacterium reads and PCR detection. PEMAX detected chemotherapy-associated changes in the microbiome otherwise missed. Our findings suggest these dyes improve characterization of the microbiome, especially chemotherapy-associated changes, and should be included in respiratory microbiome research in TB.